@sema-agent/client-core 0.98.0 → 0.100.0

package/README.md

@@ -35,7 +35,7 @@ Renamed from **`@sema-agent/wire-cc-adapter`** (0.1.x, deprecated — see *Migra
 
 ## Scope
 
-**Version:** 0.98.0
+**Version:** 0.100.0
 
 - **Today** — the adapter seam, the whole `adapt()` pipeline (all 14 A-layer arms plus the
   B/D/E tool-card layers), the notification/caps/model families, the adapter kernel (stream driver
@@ -71,12 +71,12 @@ Renamed from **`@sema-agent/wire-cc-adapter`** (0.1.x, deprecated — see *Migra
     against — the tables live upstream precisely so this package does not keep a second copy that can
     fall behind. The browser bundle really bundles the SDK through (the portability guard would
     exit 3 rather than quietly mark it external).
-  - The declared floor is `>=23.0.2` (raised from `>=21.1.0` in 0.98.0 with the sdk 23 pick-up: the live respond receipt is read as the sdk's two-arm union (applied / replayed first decision), the session rule record follows the sdk's three-bucket type, the parked-resume startup cause is the sdk's closed five-word set and the crash-converged row is the sdk's own row type; before that raised from `>=20.0.0` in 0.97.0 with the sdk 21 pick-up, together with the server and engine floors: the two decision doors' gate-kind tables (`APPROVAL_DECIDE_GATE_KINDS` / `PLAN_REVIEW_GATE_KINDS`, 21.1.0 on) are passed through from the sdk and the package's gate-kind judge is derived from them, the run terminal's failure detail reads the sdk's three-arm union and the approval card word comes from the sdk's closed card-word list; before that raised from `>=19.0.0` in 0.96.0 with the sdk 20 pick-up, together with the server and engine floors: the stream terminal reads the sdk's five-arm `done.result` union — idempotent replay body and refusal-after-open included — and fleet rows carry `workspaceSnapshot`; before that raised from `>=18.1.0` in 0.95.0 with the sdk 19 pick-up: the ask-row ending-cause table, the run-death verdict codes and the terminal-frame rule now come from the sdk's own tables; before that raised from `>=12.0.1` in 0.94.0: the request-side permission-mode vocabulary, the decision-attribution key type, the cancellation-cause list, the status-derived coarse-code list and the routing-failure denial word are now taken from the SDK itself, all of which it declares from 17.x or 18.x on; before that raised from `>=11.3.0` in 0.84.0: the per-session background listing (`sessions.background`), its capability bit `capabilities.background.listFace` and the closed background-status vocabulary are typed from 12.x on, and the package reads all three, so it no longer compiles against 11.x; before that raised from `>=11.2.1` in 0.80.0: a deny decision may now name its settler — `settledBy: "policy"` is typed on both the durable decide body and the live respond body from 11.3.0 on, and the gate record's settlement vocabulary carries its thirteenth word `policy_refused`, which this package reads to place a refusal in the policy bucket rather than the person's; the package now compiles against 11.3.0 and no consumer ships 11.2.x any more, so the older floor lost its witness; before that raised from `>=11.0.1` in 0.79.0: `capabilities.mcpProbe`, the MCP probe face (`mcpCapabilities` / `probeMcp` with `McpProbeFace`) and the write receipt's third liveness arm (`stillLive: "unknown"`) are typed from 11.2.x on, 11.2.1 adds `TaskRequest.approverPosture` and the `mandated` approval-frame key to the types and the runtime key anchor, the package now compiles against 11.2.1, and no consumer ships 11.0.x / 11.1.x any more, so the older floor lost its witness; before that raised from `>=9.8.1` in 0.78.0: `DeniedBy` carries its tenth word `read_boundary`, `rules.write` answers a `stillLive`-discriminated body, `RemovalLiveness` / `RuleWriteRequest` / `RuleWriteResult` / `RuleWriteBehavior` are exported from the SDK root and `TaskRequest.excludeAllTools` is typed from 11.x on, the package now compiles against 11.0.1, and no consumer ships 9.8.x any more, so the older floor lost its witness; before that raised from `>=9.7.1` in 0.75.0: `Capabilities.deviceExecutor.management` is typed from 9.8.x on, the package now compiles against 9.8.1, and no consumer ships 9.7.x any more, so the older floor lost its witness; before that raised from `>=9.6.0` in 0.74.0: `Capabilities.approvalsStreamLive` / `.executionLane`, `LivePendingRow.frame`, the `live_*` approval-stream events and `gates[].toolCallId` are typed from 9.7.x on, and no consumer ships 9.6.0 any more, so the older floor lost its witness; before that raised from `>=9.4.0` in 0.71.0: the `tool_disclosure` / `tool_progress` frames and `ToolApprovalFrame.readRootCandidate` are typed there; earlier: raised from `>=8.8.0` in 0.69.0: the `reasoning_end` frame and `McpStatusPanel.lastLegMcp` are typed from 9.4.0 on), and it is *witnessed*: the guard checks that an actually
+  - The declared floor is `>=23.1.0` (raised from `>=23.0.2` in 0.99.0: the package re-exports the sdk's request-attachment entry type `UserAttachment`, which the sdk declares from 23.1.0 on — a consumer on 23.0.x cannot resolve that name from the package's type declarations; runtime is unchanged; before that raised from `>=21.1.0` in 0.98.0 with the sdk 23 pick-up: the live respond receipt is read as the sdk's two-arm union (applied / replayed first decision), the session rule record follows the sdk's three-bucket type, the parked-resume startup cause is the sdk's closed five-word set and the crash-converged row is the sdk's own row type; before that raised from `>=20.0.0` in 0.97.0 with the sdk 21 pick-up, together with the server and engine floors: the two decision doors' gate-kind tables (`APPROVAL_DECIDE_GATE_KINDS` / `PLAN_REVIEW_GATE_KINDS`, 21.1.0 on) are passed through from the sdk and the package's gate-kind judge is derived from them, the run terminal's failure detail reads the sdk's three-arm union and the approval card word comes from the sdk's closed card-word list; before that raised from `>=19.0.0` in 0.96.0 with the sdk 20 pick-up, together with the server and engine floors: the stream terminal reads the sdk's five-arm `done.result` union — idempotent replay body and refusal-after-open included — and fleet rows carry `workspaceSnapshot`; before that raised from `>=18.1.0` in 0.95.0 with the sdk 19 pick-up: the ask-row ending-cause table, the run-death verdict codes and the terminal-frame rule now come from the sdk's own tables; before that raised from `>=12.0.1` in 0.94.0: the request-side permission-mode vocabulary, the decision-attribution key type, the cancellation-cause list, the status-derived coarse-code list and the routing-failure denial word are now taken from the SDK itself, all of which it declares from 17.x or 18.x on; before that raised from `>=11.3.0` in 0.84.0: the per-session background listing (`sessions.background`), its capability bit `capabilities.background.listFace` and the closed background-status vocabulary are typed from 12.x on, and the package reads all three, so it no longer compiles against 11.x; before that raised from `>=11.2.1` in 0.80.0: a deny decision may now name its settler — `settledBy: "policy"` is typed on both the durable decide body and the live respond body from 11.3.0 on, and the gate record's settlement vocabulary carries its thirteenth word `policy_refused`, which this package reads to place a refusal in the policy bucket rather than the person's; the package now compiles against 11.3.0 and no consumer ships 11.2.x any more, so the older floor lost its witness; before that raised from `>=11.0.1` in 0.79.0: `capabilities.mcpProbe`, the MCP probe face (`mcpCapabilities` / `probeMcp` with `McpProbeFace`) and the write receipt's third liveness arm (`stillLive: "unknown"`) are typed from 11.2.x on, 11.2.1 adds `TaskRequest.approverPosture` and the `mandated` approval-frame key to the types and the runtime key anchor, the package now compiles against 11.2.1, and no consumer ships 11.0.x / 11.1.x any more, so the older floor lost its witness; before that raised from `>=9.8.1` in 0.78.0: `DeniedBy` carries its tenth word `read_boundary`, `rules.write` answers a `stillLive`-discriminated body, `RemovalLiveness` / `RuleWriteRequest` / `RuleWriteResult` / `RuleWriteBehavior` are exported from the SDK root and `TaskRequest.excludeAllTools` is typed from 11.x on, the package now compiles against 11.0.1, and no consumer ships 9.8.x any more, so the older floor lost its witness; before that raised from `>=9.7.1` in 0.75.0: `Capabilities.deviceExecutor.management` is typed from 9.8.x on, the package now compiles against 9.8.1, and no consumer ships 9.7.x any more, so the older floor lost its witness; before that raised from `>=9.6.0` in 0.74.0: `Capabilities.approvalsStreamLive` / `.executionLane`, `LivePendingRow.frame`, the `live_*` approval-stream events and `gates[].toolCallId` are typed from 9.7.x on, and no consumer ships 9.6.0 any more, so the older floor lost its witness; before that raised from `>=9.4.0` in 0.71.0: the `tool_disclosure` / `tool_progress` frames and `ToolApprovalFrame.readRootCandidate` are typed there; earlier: raised from `>=8.8.0` in 0.69.0: the `reasoning_end` frame and `McpStatusPanel.lastLegMcp` are typed from 9.4.0 on), and it is *witnessed*: the guard checks that an actually
     installed SDK at that line still exports every value-level symbol this package imports and still
     declares `TaskStats.costMicroUsd` (the key `costOrNull` reads). A floor nobody ever ran is a
     promise, not a contract.
 - **Support floor:** the oldest server, engine and SDK this package still supports — server **7.129.1**,
-  engine **7.51.2**, SDK **23.0.2** (server and engine raised in 0.98.0 from 7.126.0 / 7.49.5, withdrawing support for servers 7.126.0 – 7.129.0; SDK raised in 0.98.0 from 21.1.0, with the peer floor; all three raised together in 0.97.0, from 7.125.0 / 7.48.0 / 20.0.0; raised together in 0.96.0 from 7.115.0 / 7.40.1 / 19.0.0; server / engine raised from 7.106.0 / 7.35.2 in 0.95.0 and from 7.104.0 / 7.33.8 in 0.92.0; SDK raised from 18.1.0 in 0.95.0 and from 12.0.1 in 0.94.0). The values are
+  engine **7.51.2**, SDK **23.1.0** (SDK raised in 0.99.0 from 23.0.2, with the peer floor; server and engine raised in 0.98.0 from 7.126.0 / 7.49.5, withdrawing support for servers 7.126.0 – 7.129.0; SDK raised in 0.98.0 from 21.1.0, with the peer floor; all three raised together in 0.97.0, from 7.125.0 / 7.48.0 / 20.0.0; raised together in 0.96.0 from 7.115.0 / 7.40.1 / 19.0.0; server / engine raised from 7.106.0 / 7.35.2 in 0.95.0 and from 7.104.0 / 7.33.8 in 0.92.0; SDK raised from 18.1.0 in 0.95.0 and from 12.0.1 in 0.94.0). The values are
   declared in `scripts/support-floor.json` and exported at runtime as `SUPPORT_FLOOR`; servers below the floor
   are no longer witnessed, and the readings that existed only for them were retired in the same release.
 - The direction is constitutional: **sdk (wire) + agent-types (vocabulary) → this package**,
@@ -300,24 +300,24 @@ guard still cross-checks the table by name).
 | `scripts/run-client-core-pure-test.mjs` | Consumer-view behaviour of every moved-in module; segment floors (B1–B8 + C) that only move up The non-throwing attachments verdict names a retired key, and gives its reason and way out, from the retirement rule itself; a negative control plants a second retired key in a copy of the build and checks that it is reported by its own name. The configuration refusal registry includes the inline-image refusal code, whose one sentence names no codes or identifiers, recognises only that code, and is checked against both the installed engine (minted in exactly one place) and the receiving server's older engine (not minted at all). The background-agent task-output reader hands the state and error lines that directly follow the status line to optional slots, positionally: lines written inside the agent's own result text never become slots; checked on twelve forms minted by the installed engine's own functions and by the receiving server's older engine. It also hands the result label's parenthetical note and the header bytes not mapped to a named slot to optional slots, with a rebuild invariant (status line, state, error, the unmapped header bytes, the result label and the output rebuild the unwrapped text, up to four named normalizations) checked on those same forms and on a seeded random pool of forged headers. A Bash command that is moved to the background mid-flight (by the user, or by the engine on a timeout or a queued message) is registered for the background panel by the same event and the same record as a model-requested background command: the reader recognises only the structured detached result (never the receipt prose, never the request's background flag), at most one registration is minted per card (a card both readers recognise yields one, taken from the structured result), the package keeps no record of what it has already published (the same task id on a second card is registered again, because only the consumer knows whether it still has the row), the receipt reader and the detached reader both stand down when the structured result says the engine failed to register, and the stream is compared byte for byte against the previous published build on inputs without a detached result (including two receipts for one task id), differing only by that one event on inputs that have one. The attachments env lane's product default now also carries the todo reminder (one default table shared by the absent, empty, unknown-word and `default` paths); `off` removes the key and an explicit list that leaves it out does not get it. The network word table matches terminated and other side closed as whole words: a local Unterminated string in JSON parse error is classified as internal instead of a connection failure, while the sentences undici itself produces, the errno codes and the UND_ERR family still classify as transport. The error-triage readers (`classifyTurnWireError`, `isWireTransportError` and the three sibling readers of the same fields) never throw on an error object whose `status`, `errorCode`, `message`, `cause` or `name` is a throwing getter, a revoked proxy or a throwing string conversion; an unreadable `status` is its own state (neither an HTTP answer nor a transport failure), and ordinary shapes answer exactly as a frozen copy of the previous implementation does. |
 | `scripts/run-client-core-portability-test.mjs` | Kernel / A-layer / index import closures, the runtime-dependency equality gate, barrel reachability, and a real esbuild `--platform=browser` bundle Since 0.98.0 the gate-script family sweep has four forms: a pattern that can enclose a whole block comment (judged by running it on constructed probes, not by how it is spelled), a pattern that covers the line-comment marker, a hand-written quote-pairing lexer (registration only, currently empty) and string-method calls whose first argument is a comment marker (the shared lexer library is the only home), each proved on planted spellings and clean neighbours. |
 | `scripts/run-client-core-diff-test.mjs` | Differential equivalence against the CLI reference bridge + replay-id invariant + ledger round-trip |
-| `scripts/run-seat-contract-keys-test.mjs` | The seat IPC contract: verb list ↔ SPEC ↔ types, element-wise. Since 0.83.4 the seat approval request also carries the four card bits that say whether an approval must be asked and where it came from (`mandated`, `mandate`, `ruleOffersAbsence` — the parked-row form of mandated, so the seat and the card give the same `approvalIsMandated` answer — and `origin`, which the resident-posture predicate reads), and they reach the seat through `toolPermissionRequestAskBits`, which reads them exactly as the card request does — a strict own `true`, one of the closed mandate words byte for byte, a non-empty own string — so for a host that builds the seat request through that crossing an off-list word, an empty string or an inherited key never lands on the seat; the seat validator does not become a second judge that could drop a safety ask over one bit, which leaves a host that builds the request by hand to apply the same narrow read itself. Since 0.96.0 the seat chrome list also carries the two segment-end events (`text_segment_end` and `thinking_segment_end`), so the authoritative replacement of a segment (redaction or rewrite) can reach a seat lane (since 0.98.0 also the six presentation arms the renderer already consumes: `retry_status`, `context_usage`, `last_turn_usage`, `bgshell_register`, `bgshell_settle` and `memory_saved`; `workflow_notification_enqueue` stays with the host process that owns the session's idle auto-submit and its notification queue, so the completion notice has exactly one deliverer — a real adapter-minted one is pinned as rejected by the seat gate; every `required` chrome arm must have a written seat disposition or the gate goes red, and the set of required arms may not shrink: the frozen floor is declared on the chrome arm table's own type, so turning an existing required arm optional fails to compile, and the gate checks the runtime list against that same declaration), and its type is bound to the discriminants of `ChromeEvent` itself (a word that is not a chrome arm no longer compiles); the approval request carries the seat's allow-for-this-session bit; since 0.97.0 that bit is `sessionRow` (replacing `rememberRefused`): present only when the card judge's session-row premise holds and the request id's domain can carry a session answer, absent otherwise, which hides that row — never invented as `false`; the crossing never copies the bit from its input and never mints it for a seat request itself (the seat reads the request's own bit) Since 0.98.0 it also checks that every outbound gate accepts only plain records — a Date, Map, Set, array, class instance or object with a custom prototype carrying otherwise-valid fields is refused (structured cloning would drop those fields in transit), while a plain object from another JavaScript realm is accepted — and that four unused event fields are gone from the contract. 0.98.0: the outbound seat validator lists the turn-failure superset key without ever refusing an error event because of it. |
+| `scripts/run-seat-contract-keys-test.mjs` | The seat IPC contract: verb list ↔ SPEC ↔ types, element-wise. Since 0.83.4 the seat approval request also carries the four card bits that say whether an approval must be asked and where it came from (`mandated`, `mandate`, `ruleOffersAbsence` — the parked-row form of mandated, so the seat and the card give the same `approvalIsMandated` answer — and `origin`, which the resident-posture predicate reads), and they reach the seat through `toolPermissionRequestAskBits`, which reads them exactly as the card request does — a strict own `true`, one of the closed mandate words byte for byte, a non-empty own string — so for a host that builds the seat request through that crossing an off-list word, an empty string or an inherited key never lands on the seat; the seat validator does not become a second judge that could drop a safety ask over one bit, which leaves a host that builds the request by hand to apply the same narrow read itself. Since 0.96.0 the seat chrome list also carries the two segment-end events (`text_segment_end` and `thinking_segment_end`), so the authoritative replacement of a segment (redaction or rewrite) can reach a seat lane (since 0.98.0 also the six presentation arms the renderer already consumes: `retry_status`, `context_usage`, `last_turn_usage`, `bgshell_register`, `bgshell_settle` and `memory_saved`; `workflow_notification_enqueue` stays with the host process that owns the session's idle auto-submit and its notification queue, so the completion notice has exactly one deliverer — a real adapter-minted one is pinned as rejected by the seat gate; every `required` chrome arm must have a written seat disposition or the gate goes red, and the set of required arms may not shrink: the frozen floor is declared on the chrome arm table's own type, so turning an existing required arm optional fails to compile, and the gate checks the runtime list against that same declaration), and its type is bound to the discriminants of `ChromeEvent` itself (a word that is not a chrome arm no longer compiles); the approval request carries the seat's allow-for-this-session bit; since 0.97.0 that bit is `sessionRow` (replacing `rememberRefused`): present only when the card judge's session-row premise holds and the request id's domain can carry a session answer, absent otherwise, which hides that row — never invented as `false`; the crossing never copies the bit from its input and never mints it for a seat request itself (the seat reads the request's own bit) Since 0.98.0 it also checks that every outbound gate accepts only plain records — a Date, Map, Set, array, class instance or object with a custom prototype carrying otherwise-valid fields is refused (structured cloning would drop those fields in transit), while a plain object from another JavaScript realm is accepted — and that four unused event fields are gone from the contract. 0.98.0: the outbound seat validator lists the turn-failure superset key without ever refusing an error event because of it. Since 0.99.0 the event no longer has the two positions no host ever sent (`request`, `sources`) while `code` stays as the failure event's status-code field; the event list gains `subagent_content`, whose payload is the subagent content event in exactly the package's own projection, checked end to end from a wire frame to the renderer's own content store; and the six presentation arms (retry status, context usage, last-turn usage, background shell register and settle, memory saved) carry closed payloads, with the engine's turn usage reduced to its six declared keys plus a count of the keys this package does not know. |
 | `scripts/run-approval-frame-keys-test.mjs` | The tool-approval frame key mirror, element-wise against the SDK's runtime anchor (one carve-out: AHEAD_OF_ANCHOR entries — keys the server already emits but the SDK anchor has not caught up to — may lead by one generation; the gate turns red the day the SDK catches up, forcing the entry's removal — the register is occupied again — this time by the bit that says a saved allow rule cannot retire a given approval card, carrying both the release that minted it and the byte coordinates that prove it, so the lead is a dated record rather than an exemption; its predecessor left the register the other way, by being retired upstream rather than by the anchor catching up). Beside the key mirror it now guards three further faces of that bit: the closed word table the durable leg reads it through must be **the very array object the SDK exports**, not a same-looking copy — reference identity, because an equal-contents check still permits a second table that diverges the day upstream adds a member; the one predicate a client is meant to call answers over both legs — the live card's bit and the parked row's absence word, which is all the row carries, since the row has no such bit at all — and answers `false` for a malformed value exactly as its presence-only siblings do, a strictness the upstream mint shares; and the one sentence minted for it must never point the reader at writing a rule, since a rule written in answer to a mandated question can never take effect where it was written. The same bit's key also has to reach the card request itself, which the SDK's card anchor does not list — a fact that arrives at the package boundary and stops there is the shape of defect this file's guards exist to catch. Since 0.83.2 the frame also mirrors `mandate` (the word a mandated question stands on) ahead of the SDK anchor, with its own exit condition; the released server carries it from 7.102.0, so the allowance for the server fixture was removed by its own exit condition. The installed engine must declare the member with the six-word closed type, and the key must reach the card request. Since 0.91.0 the frame also mirrors `probeReason` ahead of the development SDK anchor, with the same exit condition, and the probe-tree comparison reads the newest configured SDK first. |
 | `scripts/run-segment-authority-single-source-test.mjs` | The authoritative-segment replacement verdict, single-sourced. `text_end.content` and the `text_delta` stream stopped being byte-identical the day the engine started redacting the former through the same filter as the result, so every consumer now has to decide six ways what to do with the segment it has half-emitted — and until this release that decision existed **twice**: once here for the transcript lane, once in the shell for the print lane, hot-fixed a version apart. The verdict is now one pure function both lanes call, and the guard pins it on the quantity that actually decides the outcome: whether the authoritative text still *starts with* the bytes that already left, not whether a flush has happened — the latter is a precondition, and anchoring on it withholds a perfectly ordinary answer. Each of the six forms is checked with its counter-case, the prefix length is pinned to UTF-16 code units against a non-ASCII sample whose UTF-8 byte count differs (slicing by bytes leaves the very thing being redacted on screen), and the withheld-segment ledger is compared by normalised equality rather than substring, because a short redaction marker quoted in an unrelated later answer would otherwise suppress that answer entirely. The same file pins the session-level memory-capture declaration to one mint point — the wire value is a single-member closed set, and a consumer that spells it wrong gets a loud refusal rather than a silently dropped privacy request — and pins the SDK URL/health transit to be the **same function reference**, since wrapping it would discard the one guarantee the transit exists for. A last section strips comments with the TypeScript parser and asserts the second expression has not grown back Since 0.98.0 the gate also pins the single reading of engine liveness: no string expression in the package source evaluates, after static folding, to a `/health` path, so a second hand-written probe cannot come back however it is spelled, and the package's one liveness helper answers six real HTTP fixtures exactly as the SDK probe does (a self-reported degraded body, a non-JSON body and a 503 are not alive; 200, 201 and 299 with an ok body are). |
 | `scripts/run-print-bash-iserror-test.mjs` | The print lane's Bash `is_error` authority (structured over regex). A second section pins where the denial classification word lands on this lane: on the message envelope, never inside the tool-result block, because that block is forwarded verbatim to the provider on compaction and a self-minted key there is the shape of an old, real defect. A word outside the upstream table — or an empty string, a non-string, or nothing at all — mints no key rather than a guess, and the word never moves the error flag, because attribution does not decide anything |
 | `scripts/run-bash-benign-exit-interpretation-test.mjs` | Benign non-zero Bash exits (`returnCodeInterpretation`) stay non-errors across all three derivation arms, and the annotation transits to the card |
-| `scripts/run-sdk-floor-test.mjs` | The SDK version floor — and, more to the point, that the *installed* type declarations still carry the keys this package reads — including, from 0.80.0, the three declarations that justify the floor itself: the key naming who settled a refusal on both decision legs, and the thirteenth word in the settlement vocabulary. They are found through the syntax tree rather than by searching text, because this guard's own comment stripper blanks string contents and would have made that check permanently, silently green. From 0.84.0 the floor is 12.0.1 and the guard witnesses the declarations the package now reads: the per-session background listing method, the `background.listFace` capability bit, and the closed background-status vocabulary that the registry classification switches over exhaustively. Every installed SDK the guard can see at or above the floor must carry those declarations too, so lowering the floor to a line where they do not exist fails. From 0.94.0 the floor is 18.1.0: the shipped declarations name the approval-attribution type and the six-word request permission-mode type that SDK 17 introduced, and the cancellation-cause, status-derived-code and routing-failure values are generated from the same SDK at build time; the guard reads those values on the floor-line SDK it can see. |
+| `scripts/run-sdk-floor-test.mjs` | The SDK version floor — and, more to the point, that the *installed* type declarations still carry the keys this package reads — including, from 0.80.0, the three declarations that justify the floor itself: the key naming who settled a refusal on both decision legs, and the thirteenth word in the settlement vocabulary. They are found through the syntax tree rather than by searching text, because this guard's own comment stripper blanks string contents and would have made that check permanently, silently green. From 0.84.0 the floor is 12.0.1 and the guard witnesses the declarations the package now reads: the per-session background listing method, the `background.listFace` capability bit, and the closed background-status vocabulary that the registry classification switches over exhaustively. Every installed SDK the guard can see at or above the floor must carry those declarations too, so lowering the floor to a line where they do not exist fails. From 0.94.0 the floor is 18.1.0: the shipped declarations name the approval-attribution type and the six-word request permission-mode type that SDK 17 introduced, and the cancellation-cause, status-derived-code and routing-failure values are generated from the same SDK at build time; the guard reads those values on the floor-line SDK it can see. Since 0.99.0 the floor is SDK 23.1.0, and the floor-line witness must export the user-attachment type that the package root re-exports unchanged, with the attachment field declared on both the task request and the capabilities. |
 | `scripts/run-engine-caps-ledger-test.mjs` | A per-key disposition ledger for `GET /v1/capabilities`. The SDK's `Capabilities` grew from 74 keys to 93 in one release and nothing on the board could see it: this package consumes that table through four synchronous readers, and *nineteen new positions arriving while the package does not move* is exactly the disease shape this repo keeps logging on other axes — the fact is already on the wire, the package boundary is the cell that swallows it, and no client can read it however they write their side. So the ledger is reconciled **element-wise against the SDK interface in both directions**: a key the SDK added with no ledger row is red (someone must classify it), and a row for a key the SDK removed is red too (a registration that no longer does anything). Each row then has to survive its own claim — a `read` row names the source file, and the **code** there (comments stripped) must really mention the key, because prose asserting an alignment is the classic way these guards go hollow; a `not_read` row must have **zero** read sites in the tree, so wiring one up while the ledger still says the package ignores it is red rather than invisible. The census behind those two directions recognises five call shapes, each of which really occurs here — a reader whose base argument carries its own parentheses, a direct `caps.<key>`, a narrowing cast, an own-property read helper, and a `*_CAP` constant — and proves it on fabricated samples first, since a census that recognises one shape reports "nothing here" for the other four. What the guard deliberately does **not** judge is whether a position *ought* to be read: that is a design call, and the ledger only pins that every capability was looked at once by a person and that what they wrote down does not contradict the code In 0.98.0 the census reads the syntax tree — call heads, the second argument, constants, property reads, narrow casts and own-property reads are nodes, not text matched by a pattern with bracket balancing and quote pairing — and proves six more shapes on fabricated samples (a closing bracket inside a string, a pattern literal holding brackets and commas, a template with a closing bracket, a string with a comma and both quote kinds, an optional call, call-looking text inside a string) plus a control that removing the whole class empties the census. |
 | `scripts/run-sql-engine-capability-test.mjs` | The SQL-posture read face and the four-state capability reader underneath it. One capability cell here carries **four different things**, and each one points an operator somewhere else: nothing has been observed yet in this process (a one-shot doctor run is always in that state), the response arrived but cannot say (no store posture, or one that cannot be read), the engine reports a store posture without an SQL seat — *this deployment has no SQL backend*, which is a **positive fact** rather than an absence — and a full reading. Fold any two together and the screen states something flatly, confidently, and wrongly, so every positive control here is paired with a control pointing the opposite way, and the four sentences the doctor row can print are checked to be pairwise distinct and non-implying. The reading itself is narrowed no tighter than the mint: `txnMode: null` is a **legal value** — two of the three engines always report it that way, and the upstream type note names reading it as "optimistic" as the error — so treating it as malformed would throw away the entire reading for ordinary deployments, which is the same disease this repo logged when a consumer's domain was narrower than the producer's. A response that cannot be parsed **clears** the cell rather than leaving the previous engine's answer in place, and a separate invalidation port exists for the case the generation latch cannot catch — a same-port respawn whose new probe never succeeded, where the stale reading would otherwise be answered as current fact. Untrusted values (the isolation string is read back from a database server variable) are sanitised and bounded before display, and the bound is applied **before** escaping so a visible escape never gets cut in half. Finally the export names are themselves a guard: the shell still carries a copy that is meant to go red on the package's same-named export and be swapped out, so renaming anything here would silently disarm that lock. Current engines (≥7.106.0) moved the SQL posture under the always-present store posture (`store.sql`) and dropped the top-level key, and from 0.92.0 the reader reads `store.sql` alone (absent there means the engine reports no SQL posture): the top-level position of servers below the support floor is no longer read, a response without a readable store posture is "not reported" — the same answer the other readers sharing that witness give — and an unreadable SQL seat clears the cell; the engine's own projection is replayed to prove every form it mints reads back |
 | `scripts/run-web-search-backend-capability-test.mjs` | The deployment-default WebSearch backend read face (`capabilities.webSearch.backend`, engine ≥7.82.1). Same four-state discipline as the SQL and write-protection cells, with two things that are specific here and therefore guarded: a **missing key** (an older engine) and an explicit **`"none"`** (the engine says this deployment has no default search backend) point an operator in opposite directions — "cannot tell" versus "not configured" — and must never be folded; and the `none` sentence has to say both halves of the contract at once: the default scenario mounts no WebSearch tool, **and** a caller-supplied `webSearch` setting can still mount it on a single-user lane, because the capability advertises the deployment default, not whether this request has search. The backend word is read as an **open set** — the engine's closed set is typed from its own provider tuple and grows with it, so hand-copying three words here would turn a newly configured backend into "unreadable" (the narrower-than-the-mint disease this repo already logged once). `webSearch: null` is malformed rather than `none` (the mint never emits `null`), extra members never cross, an unparseable response clears the cell, a stale probe's late landing is dropped by the capabilities cache, the invalidation port clears to "not observed", and the open-set word is sanitised and bounded before display |
-| `scripts/run-terminal-cause-projection-test.mjs` | The `7.64.0` wire reshape, projected. A run's ending stopped being eight parallel flat keys and became **one tagged cause** (`completed \| failed \| blocked \| paused`), and a tool call's gate stopped being four orthogonal words and became **one record** (`disposition` / `settlement?` / `origin?`). Both are read in exactly one place in this package, and this guard pins them at **two levels**, because the dangerous seam is "the reader was updated, the consumer was not": each terminal arm is checked on the reader *and* on the `subtype` / `is_error` / `errors[]` the projector actually emits. Two properties carry most of the weight. First, a terminal word this reader does not know is **never** laundered into an empty success — it lands on an `unknown` arm carrying the word verbatim, while a payload with no terminal word at all (the mock lane) keeps the success arm exactly as before, which is the one and only case the reader answers `null`. Second, the three window words (`approval_window_expired`, `denial_limit_window_expired`, `park_sla_expired`) must each be told apart by a different predicate: the previous generation collapsed all three onto one `timeout`, and re-merging them would throw away the discrimination this reshape just restored. Two byte generations are read by one reader, keyed on the discriminator upstream nailed (`"terminal" in result`): the current cause form, and the **flat** form that a current engine still emits on two lanes — replayed persisted bytes, which the service passes through verbatim rather than back-filling, and the service's own rejection envelope. A cause-form payload that also carries stale flat keys must ignore them entirely: keeping one compatibility read is what gives a single fact two sources. The same file also pins the MCP delivery verdict and HTTP status riding the wiring manifest, the four-state write-protection reading (where three of the four states mean *cannot tell*, and none of them may be printed as "there is no table"), and the park-reopen fetch identity: that predicate is asserted through the **real entry point**, since the defect being fixed was precisely a call site wired to a different predicate than the one that routed the row there. From 0.80.0 one of those three boundaries flips: the key naming **who settled a refusal** stopped being a dead byte and became part of the wire, so the check stopped scanning the build output for the word and started reading the request bodies the two decision legs actually send. A refusal attributed to the deployment's own policy carries the word; one attributed to a person, one with no attribution at all, and one carrying a word the vocabulary does not hold carry nothing — the wire has no slot for “a person decided this” other than the key's absence, so inventing one would be minting a word upstream does not have. The allow family never carries it on any of its routes, because that combination is refused before the approval is judged while the side effects of allowing have already landed, and the three refusals nobody was asked about (a card that failed, a user who walked away, an interruption) carry nothing either. A deployment that signs the bodies it accepts does not sign that word, and there is no capability bit to ask beforehand, so a refusal on exactly that ground is answered by re-sending the same decision once with that one key removed — byte-for-byte the same otherwise — rather than letting an optional note take the whole denial down with it. The guard measures that along three axes: the decision still lands and is reported as decided with the attribution handed back and a separate flag saying it never reached the wire; a caller who aborted in between gets no second request; every other refusal code, and every decision that never carried the key, send exactly once. The classification of a second failure is made from what the second body actually carried, not from what the card asked for. From engine `7.104` a synchronous submit that stops at a gate returns the engine result itself plus a three-key receipt: it now carries the tagged cause, so it reads as `paused` on the cause generation (older engines still send the flat three-key body, which keeps reading as before), and the top-level park word is looked at first, the same order the SDK documents for all three generations, so a body the server says is parked is never read as finished. A parked run row now carries its result too, and the headless reconnect path turns it into the parked terminal frame on the first attempt instead of spending its whole retry budget — both generations are pinned, including an end-to-end drive through the public reconnect entry point. When the caller passes the failure's error carrier and it is a session-history tool_use mismatch, the edit-and-resend reading yields to the recovery sentence that line already carries; the carrier is handed to the same mismatch classifier, so the reading and the classifier never disagree on a carrier, whatever its shape; without a carrier the reading is compared byte for byte with the previous release's package. The machine-readable cause of a `blocked` ending is pinned as well: when a deployment hook keeps a prompt from being submitted, the blocked reading carries an optional detail with three arms — a known kind with a known cause word, a known kind with a cause word this version does not know, and a kind this version does not know — and a word outside the tables comes back verbatim and is never folded into a known one. An absent or malformed detail is left out entirely rather than guessed at, only own properties are read (planting the same names on the global prototype changes nothing), a flat-plane row and the other endings have no such position, and a reading without a detail is compared byte for byte with the previous release's package. Real engine and server packages from both generations are driven end to end — the engine's four hook endings, the server's redaction of the hook's own words, and the reader — and a per-generation fact pin says which engine and server versions mint the detail. |
+| `scripts/run-terminal-cause-projection-test.mjs` | The `7.64.0` wire reshape, projected. A run's ending stopped being eight parallel flat keys and became **one tagged cause** (`completed \| failed \| blocked \| paused`), and a tool call's gate stopped being four orthogonal words and became **one record** (`disposition` / `settlement?` / `origin?`). Both are read in exactly one place in this package, and this guard pins them at **two levels**, because the dangerous seam is "the reader was updated, the consumer was not": each terminal arm is checked on the reader *and* on the `subtype` / `is_error` / `errors[]` the projector actually emits. Two properties carry most of the weight. First, a terminal word this reader does not know is **never** laundered into an empty success — it lands on an `unknown` arm carrying the word verbatim, while a payload with no terminal word at all (the mock lane) keeps the success arm exactly as before, which is the one and only case the reader answers `null`. Second, the three window words (`approval_window_expired`, `denial_limit_window_expired`, `park_sla_expired`) must each be told apart by a different predicate: the previous generation collapsed all three onto one `timeout`, and re-merging them would throw away the discrimination this reshape just restored. Two byte generations are read by one reader, keyed on the discriminator upstream nailed (`"terminal" in result`): the current cause form, and the **flat** form that a current engine still emits on two lanes — replayed persisted bytes, which the service passes through verbatim rather than back-filling, and the service's own rejection envelope. A cause-form payload that also carries stale flat keys must ignore them entirely: keeping one compatibility read is what gives a single fact two sources. The same file also pins the MCP delivery verdict and HTTP status riding the wiring manifest, the four-state write-protection reading (where three of the four states mean *cannot tell*, and none of them may be printed as "there is no table"), and the park-reopen fetch identity: that predicate is asserted through the **real entry point**, since the defect being fixed was precisely a call site wired to a different predicate than the one that routed the row there. From 0.80.0 one of those three boundaries flips: the key naming **who settled a refusal** stopped being a dead byte and became part of the wire, so the check stopped scanning the build output for the word and started reading the request bodies the two decision legs actually send. A refusal attributed to the deployment's own policy carries the word; one attributed to a person, one with no attribution at all, and one carrying a word the vocabulary does not hold carry nothing — the wire has no slot for “a person decided this” other than the key's absence, so inventing one would be minting a word upstream does not have. The allow family never carries it on any of its routes, because that combination is refused before the approval is judged while the side effects of allowing have already landed, and the three refusals nobody was asked about (a card that failed, a user who walked away, an interruption) carry nothing either. A deployment that signs the bodies it accepts does not sign that word, and there is no capability bit to ask beforehand, so a refusal on exactly that ground is answered by re-sending the same decision once with that one key removed — byte-for-byte the same otherwise — rather than letting an optional note take the whole denial down with it. The guard measures that along three axes: the decision still lands and is reported as decided with the attribution handed back and a separate flag saying it never reached the wire; a caller who aborted in between gets no second request; every other refusal code, and every decision that never carried the key, send exactly once. The classification of a second failure is made from what the second body actually carried, not from what the card asked for. From engine `7.104` a synchronous submit that stops at a gate returns the engine result itself plus a three-key receipt: it now carries the tagged cause, so it reads as `paused` on the cause generation (older engines still send the flat three-key body, which keeps reading as before), and the top-level park word is looked at first, the same order the SDK documents for all three generations, so a body the server says is parked is never read as finished. A parked run row now carries its result too, and the headless reconnect path turns it into the parked terminal frame on the first attempt instead of spending its whole retry budget — both generations are pinned, including an end-to-end drive through the public reconnect entry point. When the caller passes the failure's error carrier and it is a session-history tool_use mismatch, the edit-and-resend reading yields to the recovery sentence that line already carries; the carrier is handed to the same mismatch classifier, so the reading and the classifier never disagree on a carrier, whatever its shape; without a carrier the reading is compared byte for byte with the previous release's package. The machine-readable cause of a `blocked` ending is pinned as well: when a deployment hook keeps a prompt from being submitted, the blocked reading carries an optional detail with three arms — a known kind with a known cause word, a known kind with a cause word this version does not know, and a kind this version does not know — and a word outside the tables comes back verbatim and is never folded into a known one. An absent or malformed detail is left out entirely rather than guessed at, only own properties are read (planting the same names on the global prototype changes nothing), a flat-plane row and the other endings have no such position, and a reading without a detail is compared byte for byte with the previous release's package. Real engine and server packages from both generations are driven end to end — the engine's four hook endings, the server's redaction of the hook's own words, and the reader — and a per-generation fact pin says which engine and server versions mint the detail. The "parked at an approval gate" predicate recognises both the live frame form and the archived session-row form. |
 | `scripts/run-auto-mode-unavailable-test.mjs` | The fact behind "you are being asked because the auto-mode classifier could not run", and the one place its sentence is minted. The cause table is a **copy**, reconciled word for word in both directions against the installed engine's own bytes — it narrowed upstream, and the guard follows rather than keeping the old shape: a table checked against something nobody ships any more is the oldest way for a guard to be green and wrong. The retirement is held from both sides — the removed table must really be gone upstream, and the removed reader and word must really be gone here — while the word that left keeps arriving cleanly from an older engine, because the reader takes the cause as an **open set**: the vocabulary belongs upstream, so a copied list here would discard a legal value the day one is added, and the value discarded is precisely "this outage is a NEW kind". The reader's one exclusion is the word the engine says it never stamps here — the classifier did run and did answer, just outside its contract, so reading it as a failure would invent an event the engine denies. That exclusion used to be derived from a second table which no longer exists; the reason for it never lived in that table, so it is now stated where it actually comes from, pinned as a **named** set (a magic literal scattered through the reader reds) and cross-checked against the engine's own verdict declaration and against the reader having exactly one such comparison. One reader serves both the live ask and its durable parked twin, since the two carry the same key path and a second copy is how two ledgers drift apart. Absence is pinned as absence — most asks never consulted a classifier at all — and the sentences are checked mutually distinct, prototype-safe, and walked end to end: an unknown word reaches the sentence a person reads (the fallback that names it verbatim) and the status reading (unavailable for this round, never a fallback to "available"), with counter-controls proving neither assertion is vacuous |
-| `scripts/run-engine-notice-catalog-test.mjs` | The engine-notice catalog and its audience table. Whether a notice deserves a person's attention is not decided by whether this end happens to have a phrasing for it — that drifts with each client's build order — but by whether the engine minted the code into its own written catalog; the audience row answers the separate question of *who* the fact is for, since an operations fact pushed at an end user is noise and a user-facing fact buried in an operator log is something withheld from the person who could act on it. Both tables are reconciled against the installed engine's own artefacts in both directions and pinned in lockstep with each other, unknown codes fall back to the conservative operator side, and catalog membership is tested on the raw value so a code carrying control characters cannot impersonate a registered one after sanitizing. The reader for a dropped MCP injection keys on its own code alone and treats a missing session, server or reason as absence rather than throwing at a read site. A reverse pin enforces the upstream's single-mint contract: the engine composes those sentences from the host's facts, so a copy of them appearing in this package's source or build is a second source that would drift, and fails. From 0.84.0 it also covers the reader for the two read-directory grant notices: it recognises only those two codes, needs the tool call id to match a card, passes the rejection reason through as written, and treats only the granted notice as evidence that a directory was added; a granted notice without both the directory and the spelling the engine now holds, or with a scope other than `exact`, is not read at all, and the scope word is pinned to the engine's type at compile time. The server also mints a few notices of its own through the same channel; those codes live in a second table with their own audiences, kept apart from the engine mirror (which must stay equal to the engine's catalog) and reconciled against the server's published package when one is supplied, so a user-facing server notice is no longer filed under operations. One dispatcher returns the typed facts for every code that has a reader, discriminated by code and tagged with its audience — only the user-audience codes belong on a user surface — and the guard ties the dispatch table to the module's own exported readers in both directions, so a reader cannot be exported without a row and a row cannot be dropped without the guard failing. From 0.85.0 it also covers a third server-minted notice, the one saying that part of a session's saved history could not be read when the session was reopened: it is a user-audience notice, its two counts are read one by one and anything that is not a non-negative integer reads as `unknown` rather than zero (neither "nothing was skipped" nor "nothing is left" may be invented), and one extra sentence — the context is empty but the turn runs — is given only when the count of entries left is exactly zero. When the server package is supplied, the guard drives the server's own emitter for that notice and reads what it emits back through the dispatcher. From 0.86.0 the catalog grows by six engine notices — a personal rule store that could not be read and was skipped, an auto-mode classification that could not decide and where the call went, a requested permission mode that did not take effect, a permission mode answering a question on the person's behalf, a mode release past a read-deny table, and a legacy checkpoint's mode keys being migrated — each with its own typed reader: branching words (the classifier's cause and destination, the reason a mode did not take effect) are read as closed sets, descriptive words pass through as written, and a malformed notice reads as absent; two of them are checked against notices produced by the engine's own emitters. The unresolvable-ask notice's optional remedy sentence is carried verbatim when it is a non-empty string — never trimmed, rewritten or filtered by cause — and its absence (older engines, other causes, an empty or non-string value) leaves the rest of the view unchanged; a notice without it reads byte-for-byte as before, and the engine's own mint function drives three cases through the dispatcher. A fourth server-minted code, the user-facing notice that a model endpoint could not be connected to, is registered with its audience and has its own reader: the failure code, the endpoint, the remedy sentence and the session are required, the name of the proxy variable is optional, the code and remedy pass through as written, and the server's sentences are never copied into the package. With server fixtures, the server's own decorator mints the notice for a direct and a proxied connection and for every connect-failure code it knows, and stays silent when a response arrived or the call did not end in error; the audience table is reconciled against both fixture generations, a row newer than the older fixture being required to be absent there. From 0.88.0 the catalog grows by one operator notice — a provider request whose tool turn broke the pairing rules was repaired on its way out — with a typed reader: each repair's form word passes through as written (the engine does not publish that word list from its entry point and has announced more words), the call ids are read as a non-empty list, any malformed entry makes the whole notice read as absent, and notices produced by the engine's own repair law and emitter read back field for field. The model-unreachable reader reads only the notice detail's own properties, like the other server-minted readers. From 0.89.0 the engine's own catalog registers those server-minted codes as well — each code carries an audience and a record of who mints it — so the package's separate table for them is retired and both the membership check and the audience lookup answer from the engine's tables alone, the four codes keeping the audiences they had; the guard reconciles the engine's record of server-minted codes against the published server's stream allow-list on the newer server generation and against the older server's own table on the older one, and the generator refuses an engine whose minter table is not in lockstep with its catalog. Two user-facing notices gain typed readers: an incomplete tool turn (cause, count, session and run are required, the call ids are optional and copied, and a cause word the package does not know passes through as written) and user input a run may not have used — one reader for the steer and follow-up codes, the kind decided by the code, the count required, the cause passed through as written and never filled in when absent — plus one user-facing sentence that names only the count, does not claim the input went unused when the server could not tell, and falls back to a cautious sentence for any cause it does not recognise. Notices produced by the engine's and the server's own emitters read back field for field, and every reader answers the same for frozen and unfrozen input. Every reader and the by-code entry point read a notice's code and detail as own properties only: a notice whose code or detail sits on its prototype, or on a polluted global prototype, reads as nothing through all three doors alike. From 0.90.0 the catalog is eighty-five notices: the engine retired the notice for the one-off migration of old parked rows, so the catalog, audience and minter tables lose a row, its fact reader and named type are gone from the public surface, and the code is treated like any code outside the catalog (the dispatcher answers nothing, the catalog check answers false, the audience stays the conservative default). The unused-input reader also reads the child-session field carried by the copy a parent session receives when a sub-agent ends with inputs unused — taken from the engine's real minting site — and the user-facing sentence says so for the two definite shapes, with no identifier in it; every malformed form of the field drops only that field. The register now holds the allow-list notice under its 0.92.0 engine name with the user audience; the earlier engine name is not registered and falls to the operator default, which is the audience it had on the engines that minted it. Since 0.94.0 the user-input-not-used reader takes its lane and count key from the generated lane table rather than from two hand-written codes, and the catalogue expectations follow the installed engine generation. From 0.94.0 the allow-list notice has a fact reader: each entry that reads as nothing comes back with its list source, its machine value exactly as the server sent it (control characters included), a display form produced by the package's single untrusted-text exit, its reason and the optional skill, list, verdict, current-name and value-type positions; a malformed entry list makes the whole reading absent. The model-unreachable reader carries the leg's run id as an optional `taskId` when the server sends one, and drops only that key when it is unusable. Since 0.96.0 the dropped-tool-turn reader also carries the optional outcome word (absent means the engine did not say), and the catalog follows the renamed path-view notice and three new route notices. |
+| `scripts/run-engine-notice-catalog-test.mjs` | The engine-notice catalog and its audience table. Whether a notice deserves a person's attention is not decided by whether this end happens to have a phrasing for it — that drifts with each client's build order — but by whether the engine minted the code into its own written catalog; the audience row answers the separate question of *who* the fact is for, since an operations fact pushed at an end user is noise and a user-facing fact buried in an operator log is something withheld from the person who could act on it. Both tables are reconciled against the installed engine's own artefacts in both directions and pinned in lockstep with each other, unknown codes fall back to the conservative operator side, and catalog membership is tested on the raw value so a code carrying control characters cannot impersonate a registered one after sanitizing. The reader for a dropped MCP injection keys on its own code alone and treats a missing session, server or reason as absence rather than throwing at a read site. A reverse pin enforces the upstream's single-mint contract: the engine composes those sentences from the host's facts, so a copy of them appearing in this package's source or build is a second source that would drift, and fails. From 0.84.0 it also covers the reader for the two read-directory grant notices: it recognises only those two codes, needs the tool call id to match a card, passes the rejection reason through as written, and treats only the granted notice as evidence that a directory was added; a granted notice without both the directory and the spelling the engine now holds, or with a scope other than `exact`, is not read at all, and the scope word is pinned to the engine's type at compile time. The server also mints a few notices of its own through the same channel; those codes live in a second table with their own audiences, kept apart from the engine mirror (which must stay equal to the engine's catalog) and reconciled against the server's published package when one is supplied, so a user-facing server notice is no longer filed under operations. One dispatcher returns the typed facts for every code that has a reader, discriminated by code and tagged with its audience — only the user-audience codes belong on a user surface — and the guard ties the dispatch table to the module's own exported readers in both directions, so a reader cannot be exported without a row and a row cannot be dropped without the guard failing. From 0.85.0 it also covers a third server-minted notice, the one saying that part of a session's saved history could not be read when the session was reopened: it is a user-audience notice, its two counts are read one by one and anything that is not a non-negative integer reads as `unknown` rather than zero (neither "nothing was skipped" nor "nothing is left" may be invented), and one extra sentence — the context is empty but the turn runs — is given only when the count of entries left is exactly zero. When the server package is supplied, the guard drives the server's own emitter for that notice and reads what it emits back through the dispatcher. From 0.86.0 the catalog grows by six engine notices — a personal rule store that could not be read and was skipped, an auto-mode classification that could not decide and where the call went, a requested permission mode that did not take effect, a permission mode answering a question on the person's behalf, a mode release past a read-deny table, and a legacy checkpoint's mode keys being migrated — each with its own typed reader: branching words (the classifier's cause and destination, the reason a mode did not take effect) are read as closed sets, descriptive words pass through as written, and a malformed notice reads as absent; two of them are checked against notices produced by the engine's own emitters. The unresolvable-ask notice's optional remedy sentence is carried verbatim when it is a non-empty string — never trimmed, rewritten or filtered by cause — and its absence (older engines, other causes, an empty or non-string value) leaves the rest of the view unchanged; a notice without it reads byte-for-byte as before, and the engine's own mint function drives three cases through the dispatcher. A fourth server-minted code, the user-facing notice that a model endpoint could not be connected to, is registered with its audience and has its own reader: the failure code, the endpoint, the remedy sentence and the session are required, the name of the proxy variable is optional, the code and remedy pass through as written, and the server's sentences are never copied into the package. With server fixtures, the server's own decorator mints the notice for a direct and a proxied connection and for every connect-failure code it knows, and stays silent when a response arrived or the call did not end in error; the audience table is reconciled against both fixture generations, a row newer than the older fixture being required to be absent there. From 0.88.0 the catalog grows by one operator notice — a provider request whose tool turn broke the pairing rules was repaired on its way out — with a typed reader: each repair's form word passes through as written (the engine does not publish that word list from its entry point and has announced more words), the call ids are read as a non-empty list, any malformed entry makes the whole notice read as absent, and notices produced by the engine's own repair law and emitter read back field for field. The model-unreachable reader reads only the notice detail's own properties, like the other server-minted readers. From 0.89.0 the engine's own catalog registers those server-minted codes as well — each code carries an audience and a record of who mints it — so the package's separate table for them is retired and both the membership check and the audience lookup answer from the engine's tables alone, the four codes keeping the audiences they had; the guard reconciles the engine's record of server-minted codes against the published server's stream allow-list on the newer server generation and against the older server's own table on the older one, and the generator refuses an engine whose minter table is not in lockstep with its catalog. Two user-facing notices gain typed readers: an incomplete tool turn (cause, count, session and run are required, the call ids are optional and copied, and a cause word the package does not know passes through as written) and user input a run may not have used — one reader for the steer and follow-up codes, the kind decided by the code, the count required, the cause passed through as written and never filled in when absent — plus one user-facing sentence that names only the count, does not claim the input went unused when the server could not tell, and falls back to a cautious sentence for any cause it does not recognise. Notices produced by the engine's and the server's own emitters read back field for field, and every reader answers the same for frozen and unfrozen input. Every reader and the by-code entry point read a notice's code and detail as own properties only: a notice whose code or detail sits on its prototype, or on a polluted global prototype, reads as nothing through all three doors alike. From 0.90.0 the catalog is eighty-five notices: the engine retired the notice for the one-off migration of old parked rows, so the catalog, audience and minter tables lose a row, its fact reader and named type are gone from the public surface, and the code is treated like any code outside the catalog (the dispatcher answers nothing, the catalog check answers false, the audience stays the conservative default). The unused-input reader also reads the child-session field carried by the copy a parent session receives when a sub-agent ends with inputs unused — taken from the engine's real minting site — and the user-facing sentence says so for the two definite shapes, with no identifier in it; every malformed form of the field drops only that field. The register now holds the allow-list notice under its 0.92.0 engine name with the user audience; the earlier engine name is not registered and falls to the operator default, which is the audience it had on the engines that minted it. Since 0.94.0 the user-input-not-used reader takes its lane and count key from the generated lane table rather than from two hand-written codes, and the catalogue expectations follow the installed engine generation. From 0.94.0 the allow-list notice has a fact reader: each entry that reads as nothing comes back with its list source, its machine value exactly as the server sent it (control characters included), a display form produced by the package's single untrusted-text exit, its reason and the optional skill, list, verdict, current-name and value-type positions; a malformed entry list makes the whole reading absent. The model-unreachable reader carries the leg's run id as an optional `taskId` when the server sends one, and drops only that key when it is unusable. Since 0.96.0 the dropped-tool-turn reader also carries the optional outcome word (absent means the engine did not say), and the catalog follows the renamed path-view notice and three new route notices. Since 0.99.0 the catalog follows the newer engine at 95 codes, and the audience of each new code is the engine's own. |
 | `scripts/run-tool-roster-projection-test.mjs` | The leg's tool roster — what the engine says it actually mounted and what face each tool wears — replacing three word lists that were only ever an estimate taken from one traffic capture against one pinned engine. The reader copies the engine's own all-or-nothing discipline: a roster whose row cannot be read, or whose declared count disagrees with the rows, is dropped whole rather than handed over short, because a consumer reading a short roster concludes the missing tools are not mounted — the upstream says in as many words that this is worse than sending nothing. A malformed *face* on a row (path target, render hints) drops only that face, since a face is not an identity. Shims are built strictly from roster rows and never guessed from a tool's name, and an axis that cannot be read stays absent rather than defaulting to `false` or `never`, which would render "unknown" as "safe". For run-time changes the guard pins the one hard rule in the contract: a digest that does not match is **not** a rejection — the carried roster is the new state regardless and only the summary becomes unusable, because refusing the swap would leave the consumer holding a stale roster forever. One reading here answers a question that the terminal state structurally cannot: whether this run was assembled with any file-and-shell tools at all. The engine's terminal vocabulary says a run finished, not whether the work got done, so an orchestrator that waits for the end and then guesses has nothing to guess from — while the assembly manifest already said it at the start, one row per mounted instance with the single condition that mounted it. The reading is three-state and both folds are refused: a roster that is readable and carries no such row is the engine stating a fact, while no roster at all is not that fact — the static half of a manifest never carries one, and an older engine reports rosters without naming the mount condition at all, where an empty count would be a statement about the reader rather than about the run. Those two are kept apart in the reason the reading carries, and the wording for every unknown case is checked never to claim the run had no tools. The same roster now decides the tool list on the first line of a non-interactive run: the host holds that line until the roster arrives and lists exactly what the engine mounted at the start of the run, in mount order. The guard runs a real assembly frame through the projection into the decision, and pins that the host falls back to the estimate only once the roster is known not to be coming — a manifest without one, an unreadable one, model output or the run's end arriving first — rather than on a timer alone (model activity counts, including a model call that is still waiting or retrying; an error line the stream synthesizes when a run fails before assembly counts as the run ending), that a sub-run's manifest is never mistaken for the run's own, that an empty roster is taken as the engine's answer rather than as silence, and that the wait bound covers both sequential default budgets the engine gives an external tool server to connect and list its tools. The holding logic itself lives in the package as a small per-run gate — buffer, decide once, release the held messages in arrival order, then pass through — and the guard drives real stream output through it to pin that the release happens exactly once, at the manifest, releasing exactly the held prefix. The ordering itself also lives in the package as a stream wrapper, and the guard checks the final output a consumer reads: the first line is always the tool-list line, a message that arrives while that line is still being built comes after it, a timer firing races nothing out of order, a source that ends or fails before the decision still gets its first line and held messages out before the error, and an early exit closes the source. From 0.84.0 the roster-derived sentence source no longer throws on a value it does not recognise, including a reading of the manifest's `hands` section passed by mistake: it answers the same "not stated" sentence as the `hands` reader, from one shared source, and its six known sentences do not change. From 0.86.0 that first line also says where its tool list came from, in two added keys the wrapper writes onto the host's own line object in place: a source of `engine-roster`, `estimate` or `host-static`, and, only for an estimate, which of the four reasons made the roster known not to be coming. The guard checks every decision path end to end, that the host's object keeps its identity and every other key byte for byte, that stale values of the same two keys left by the host are replaced (a stale reason is removed rather than left behind), and that a host line that cannot be written is passed through untouched instead of failing the first line. |
-| `scripts/run-permission-rule-issue-codes-test.mjs` | The rule-lint refusal codes an engine reports when it will not compile a permission rule. The SDK publishes neither a schema nor a type for them, so the package mints the table from the engine's own bytes and the guard pays the cost of that copy instead of leaving it to somebody remembering: it parses the codes the engine actually mints and reconciles them against the table in both directions, so a code added upstream (the user would see a bare code) and a code only the package believes in (a branch that can never fire) both fail. It also reconciles the table plus a small retired ledger against the engine's declared union, which is deliberately not the same set — one member was renamed and its old name is still declared — so reviving a code the engine will never mint again is impossible and a future stale member shows up immediately. Sentences are pinned one per code, mutually distinct, and split by family: a rule that is wrong and a rule that is legal but unsupported on this lane are different next steps and may not share a sentence. The engine's own message rides along as prose — sanitized and capped after escaping, never matched on The set of codes the engine actually mints now also counts declared codes minted outside the rule compiler (the allow layer reports an entry that is not a string with one). Codes an engine release retired but older engines still mint are found by comparing the receiving server's bundled engine with the installed one rather than by a hand-kept list; each keeps the exact sentence it had before, and the codes that replaced them each get their own sentence with no machine identifiers in it. Since 0.94.0 the base-path refusal sentence is checked to hold on both engine generations (base missing, or declared but not a usable absolute path), with the premise run on the installed engine. |
+| `scripts/run-permission-rule-issue-codes-test.mjs` | The rule-lint refusal codes an engine reports when it will not compile a permission rule. The SDK publishes neither a schema nor a type for them, so the package mints the table from the engine's own bytes and the guard pays the cost of that copy instead of leaving it to somebody remembering: it parses the codes the engine actually mints and reconciles them against the table in both directions, so a code added upstream (the user would see a bare code) and a code only the package believes in (a branch that can never fire) both fail. It also reconciles the table plus a small retired ledger against the engine's declared union, which is deliberately not the same set — one member was renamed and its old name is still declared — so reviving a code the engine will never mint again is impossible and a future stale member shows up immediately. Sentences are pinned one per code, mutually distinct, and split by family: a rule that is wrong and a rule that is legal but unsupported on this lane are different next steps and may not share a sentence. The engine's own message rides along as prose — sanitized and capped after escaping, never matched on The set of codes the engine actually mints now also counts declared codes minted outside the rule compiler (the allow layer reports an entry that is not a string with one). Codes an engine release retired but older engines still mint are found by comparing the receiving server's bundled engine with the installed one rather than by a hand-kept list; each keeps the exact sentence it had before, and the codes that replaced them each get their own sentence with no machine identifiers in it. Since 0.94.0 the base-path refusal sentence is checked to hold on both engine generations (base missing, or declared but not a usable absolute path), with the premise run on the installed engine. Since 0.99.0 the tool-name refusal code minted by the newer engine is in the table with its own sentence, and the two-way equality with the engine's minted set covers it. |
 | `scripts/run-gate-vocabulary-test.mjs` | The two gate vocabularies — who denied a call (`DeniedBy`, the installed engine's ten words) and who asked about it (`AskOrigin`, eleven) — together with the one place their sentences are minted, so the same denial does not read three different ways across three clients. The tables are copies, not opinions: the asker table is reconciled in both directions against the lists the installed SDK and engine packages publish, and the denier table is exactly the installed engine's own list (from 0.92.0 the one word only engines below the support floor still sent is no longer in it), checked so that every word the SDK declares is present except that retired word, and every word ahead of the SDK comes from the engine — a word added upstream (nobody renders it, the user sees a bare code) and a word only the package believes in (a branch that can never fire) both fail. Every word must carry its own literal sentence and no two may collide, including the sibling pairs the upstream deliberately split apart — an organization store and a personal rule store being unreadable send you to different people, and the two tighten origins exist precisely to name which layer of engine logic asked. The two fallbacks are pinned distinct because an unknown word means different things in each: for a denial layer this build does not know, the sentence says only that this build cannot say which layer refused the call or why — it no longer guesses at a newer engine or a damaged record, since the most likely unknown word is the retired one replayed from an old session; the asker vocabulary is genuinely open (the server only checks for a non-empty string, so an unknown word just means the client is older than the engine). Alongside them sits an **uplift anchor** rather than a third table: the reason a call was decided the way it was is a distinct semantic face from who denied it and who asked, one upstream has not mirrored into the SDK at all, and one whose newest member — a shell command allowed because it only reads — has no sentence anywhere yet. Minting the union here would create the second drifting source the day upstream publishes it, so the guard instead asserts the **absence** from both ends: the SDK declarations carry no such union near that word, and the installed engine’s own list does not carry the word either. The engine end fires first, on the batch that raises the dependency, which is exactly when the ownership question should be answered; the SDK end fires when the mirror lands. Either red is the work order to mint the sentence, never a reason to delete the anchor. A fourth mint now sits beside the three tables and is not a table at all: a single presence-only fact — that no saved rule and no standing posture can retire this question — earns one sentence, taking no argument precisely so a caller cannot mistake it for a second kind of mandate, pinned distinct from every sentence the tables mint, pinned never to point at rule-writing, and pinned not to overclaim the stronger neighbouring demand that a person rather than a configuration must answer; it must not say the question is asked every time — an answer for this one call may come from the person, a hook or an automatic check the deployment runs — and its wording is checked against the engine package's own description of the mandate. A fifth table joins them from 0.80.0: the words for **how a wait ended** (thirteen from 0.80.0, fourteen from 0.90.0), mirrored in both directions from the engine's own declarations — the table's owner — with the wire SDK's copy held alongside as a second witness that must match it word for word and in order, so the day the SDK falls a generation behind, that is what turns red rather than the mirror silently following the wrong source. The newest of them says a deployment's own policy answered the card — not a person, and not “nobody could be asked” — so the guard pins it apart from both neighbours by behaviour, feeding every one of the words through all five named predicates and checking which word makes which one speak, rather than what any predicate returns. Two of them also decide how a refusal is filed in the session transcript; that mapping is minted once and reused by both of the package's own entry points, and anything outside those two words yields nothing rather than a guess. Since 0.83.2 a sixth list covers the word a mandated question stands on (`APPROVAL_MANDATE_WORDS`, six words): it must equal the engine's own list word for word and in order, membership is exact, the card reader `readApprovalMandate` answers only for an own key holding one of the six words, and each word has one fixed sentence explaining why the question must be confirmed — six distinct sentences that never point the reader at writing a rule, never promise a question every time, never claim only a person may answer, and repeat no other sentence the package mints. The list is also pinned against the engine's type at compile time in both directions, while the published build references no engine package at all: every `.js` and `.d.ts` file in the build is scanned, and the same scan is first shown to fire on references planted in a scratch directory. From 0.90.0 the settlement table carries `policy_allowed` (a machine allow, listed before `policy_refused`) and the mirror is checked against the installed engine byte for byte, in order; the installed SDK (12.x) predates the word, so its list is compared as "the engine's list minus one registered lagging word", and every SDK witness at 16.0.0 or later must carry exactly the engine's set of words. The party that settled is the engine's five-word set, read open (any word, including an unknown one, comes back as written); a policy allow is no person's approval and no refusal (all five named predicates stay silent on it, and it files no denial class in the transcript), and the refusal word reads the same under both party spellings of the two server generations. The one lagging word that loosens the compile-time check against the installed SDK is itself guarded: it must stay a non-empty union of string literals (widening it to `string` compiles cleanly, so only this check catches it), the check's declared type must stay exactly as written, the word list must equal the list this check registers, and every lagging word must be absent from the SDK at the peer floor yet present in the newest witness SDK and in the installed engine's list. The origin word on a gate record and on a released-by-mode record is read as an own key only, and a structural pin holds that every source file reading an ask origin goes through the one reader. A retired mandate word that an older server still sends reads as the sentence of the word that replaced it (one sentence per fact), the sentence table holds the current words only (a leftover sentence for a retired word fails to compile), and the modules that read the rename table are pinned to a named roster. |
 | `scripts/run-engine-identity-test.mjs` | The engine generation anchors on `/health` (`pid`, `instanceId`, `startedAt`; engine >=7.67.0). `/health` is the one unauthenticated door and its heartbeat is always green, so "another host restarted the shared engine" used to be discoverable only by having some authenticated request hit a 401 first — a path that misreads a restart as a network fault. The reader narrows each anchor independently (one malformed field never hides the other two) and always hands back a reading object rather than an absence, because the caller is asking which anchors answered, not whether there was a response. The comparison is a three-word verdict, not a boolean: `unknown` when the two readings share no comparable anchor at all — an empty intersection means nothing could be compared, never that nothing changed — and the boolean convenience is pinned so that only `true` is an assertion. Any comparable anchor differing decides `changed`, so a reading whose `startedAt` matches while its `instanceId` does not cannot be waved through as the same life; precedence only decides which anchor gets named in the diagnosis |
 | `scripts/run-posture-knob-projection-test.mjs` | The three deployment knobs on the operator face (`serverGates.durableApproval` / `streamAskWindowMs` / `sessionAutoTitle`, engine >=7.67.0), each read as a value **plus who set it plus one operator-facing pointer** rather than a bare value — a bare boolean cannot answer why this particular machine is on this setting or how to pin it back, and a default that flips with the deployment shape is invisible without that. A worker too old to report readings still sends a bare boolean; the reader folds it into the same shell so consumers keep one branch, but raises a `legacy` bit, answers `undefined` from the machine-readable source accessor, and mints a sentence that contains no source word at all — claiming a source nobody reported is worse than admitting the worker cannot say. The other two knobs are honestly absent on such a worker rather than defaulted, a malformed side knob drops only itself while the anchor knob drops the whole reading, and the four sentences are pinned literally distinct so an operator can tell "not observed" from "not reported" from a real value. The last leg reads the installed SDK's `openapi.yaml` and `types.d.ts` directly, including a pin that exactly one knob on this face is numeric — the premise the millisecond-to-prose rendering rests on |
-| `scripts/run-terminal-facts-projection-test.mjs` | The four unconsumed terminal-receipt facts: `TaskResult.effectiveReasoning` / `effectiveMemoryScopes` are narrowed into `_sema_effective_reasoning` / `_sema_effective_memory_scopes` on the CC-shaped `result` (success and error envelopes alike; a malformed value mints nothing, never a default tier), the resume **reopen** family (`resume.env_failed` / `tool_unavailable` / `tool_contract_mismatch`) is a frozen closed set with a reader and three-sentence copy that is disjoint from the refusal and retry-later sets, and `routePairingVerdict` reads `ModelInfo.routePairing` as ok / broken / unknown without policing the open set. A fifth section pins the structured-output key on the success result: the CC-spelled `structured_output` is the only home for the value the wire calls `structuredOutput`. The camelCase spelling this package used to mint on its own — a misspelling of the CC field, not an additive field of our own — rode alongside it for exactly one release (0.79.1) and is **absent from 0.80.0 on**, pinned both by own-key and by `in`, so a consumer still reading the old name sees `undefined` rather than a stale copy. The wire position is read exactly once, so a value-changing accessor is only ever asked for its first answer; absence stays absence; a wire key that is present but `undefined` mints nothing, since a key whose value is `undefined` makes a consumer that tests presence read "the engine produced nothing" as "the engine produced an empty result"; falsy-but-present values such as `null`, `0`, `""` and `false` are still minted, and so are shapes that are not records at all — an empty array, a populated array, a string, a number, a boolean — each carried through by the same reference, because the shape of that value is decided by the caller's own schema and the package does not get to filter it; and the error envelope carries no such key, because the CC error arm has no such field. Which spelling CC itself declares is witnessed from the mirror's own syntax tree rather than a constant copied into the guard, so the day that field is renamed upstream the guard says so. A seventh section covers the two keys the memory-scope reading used to leave behind: the mounted arm's `projectKey` (exactly one of `{ scope }` or `{ absent }`; a missing key means the deployment declared nothing, which is not the same as having looked and found none, and the reason word is read as an open set so a word added upstream is never dropped) and each row's `shadowed` (a positive integer only, never on the write home's own row, never a zero or a coerced value). A malformed or unreadable value mints that one key and nothing else, so a throwing accessor never takes the arm down; the other arms never carry either key; and inputs without the two keys read byte for byte as the previous release does, checked against that release's published artifact on a large generated pool and on a pool of live objects (throwing accessors, revoked proxies) where even throwing versus not throwing must match. The upstream facts are pinned on every generation of server fixture the guard is given: the mint sites and declarations in the engine each one bundles, the server's terminal shaping run for real so both keys cross the wire unchanged, and the server's own project-identity minting fed into the reader. The SDK types are pinned as not modelling the face at all, and the two keys that were declined are pinned as having no mint site on the wire, so the day either changes the guard goes red instead of the decision going stale. The key set compared across the success and error envelopes is read from the type declaration of the shared fact parts by the compiler's checker rather than from a hand-kept list or a name prefix: every declared key must have an input fixture and every fixture a declared key, and a planted extra key or a removed key is shown to change the derived set. |
+| `scripts/run-terminal-facts-projection-test.mjs` | The four unconsumed terminal-receipt facts: `TaskResult.effectiveReasoning` / `effectiveMemoryScopes` are narrowed into `_sema_effective_reasoning` / `_sema_effective_memory_scopes` on the CC-shaped `result` (success and error envelopes alike; a malformed value mints nothing, never a default tier), the resume **reopen** family (`resume.env_failed` / `tool_unavailable` / `tool_contract_mismatch`) is a frozen closed set with a reader and three-sentence copy that is disjoint from the refusal and retry-later sets, and `routePairingVerdict` reads `ModelInfo.routePairing` as ok / broken / unknown without policing the open set. A fifth section pins the structured-output key on the success result: the CC-spelled `structured_output` is the only home for the value the wire calls `structuredOutput`. The camelCase spelling this package used to mint on its own — a misspelling of the CC field, not an additive field of our own — rode alongside it for exactly one release (0.79.1) and is **absent from 0.80.0 on**, pinned both by own-key and by `in`, so a consumer still reading the old name sees `undefined` rather than a stale copy. The wire position is read exactly once, so a value-changing accessor is only ever asked for its first answer; absence stays absence; a wire key that is present but `undefined` mints nothing, since a key whose value is `undefined` makes a consumer that tests presence read "the engine produced nothing" as "the engine produced an empty result"; falsy-but-present values such as `null`, `0`, `""` and `false` are still minted, and so are shapes that are not records at all — an empty array, a populated array, a string, a number, a boolean — each carried through by the same reference, because the shape of that value is decided by the caller's own schema and the package does not get to filter it; and the error envelope carries no such key, because the CC error arm has no such field. Which spelling CC itself declares is witnessed from the mirror's own syntax tree rather than a constant copied into the guard, so the day that field is renamed upstream the guard says so. A seventh section covers the two keys the memory-scope reading used to leave behind: the mounted arm's `projectKey` (exactly one of `{ scope }` or `{ absent }`; a missing key means the deployment declared nothing, which is not the same as having looked and found none, and the reason word is read as an open set so a word added upstream is never dropped) and each row's `shadowed` (a positive integer only, never on the write home's own row, never a zero or a coerced value). A malformed or unreadable value mints that one key and nothing else, so a throwing accessor never takes the arm down; the other arms never carry either key; and inputs without the two keys read byte for byte as the previous release does, checked against that release's published artifact on a large generated pool and on a pool of live objects (throwing accessors, revoked proxies) where even throwing versus not throwing must match. The upstream facts are pinned on every generation of server fixture the guard is given: the mint sites and declarations in the engine each one bundles, the server's terminal shaping run for real so both keys cross the wire unchanged, and the server's own project-identity minting fed into the reader. The SDK types are pinned as not modelling the face at all, and the two keys that were declined are pinned as having no mint site on the wire, so the day either changes the guard goes red instead of the decision going stale. The key set compared across the success and error envelopes is read from the type declaration of the shared fact parts by the compiler's checker rather than from a hand-kept list or a name prefix: every declared key must have an input fixture and every fixture a declared key, and a planted extra key or a removed key is shown to change the derived set. Since 0.99.0 the route-pairing verdict table is pinned at compile time in both directions to the engine's pairing-status type, the two new broken states read as broken, and anything outside the known words, prototype key names included, reads as unknown. |
 | `scripts/run-export-liveness-test.mjs` | Every runtime export in the public baseline must be **alive**: referenced by some gate, or explicitly registered in `scripts/export-liveness.json` as `contract` (consumed by a client with no gate yet), `internal` (an internal helper amplified onto the public surface by `export *`), `candidate` (with ticket + retire-by) or `retire` (dead; retire-by version). Registration is accounting, not exemption: a row for a name a gate already references is stale and must go, a row for a name no longer exported is red, `retire`/`candidate` rows go red the moment `package.json` reaches their retire-by version, and the row count only ratchets down. When the client repositories are on disk the consumption evidence is checked by name, from committed history and never from a working tree: each client's `origin/main` (or its HEAD when there is no such ref), the checked-out HEAD however old it is, and every local branch head committed within the last 14 days, so an import that so far exists only on an unmerged branch still blocks the retirement. Two readings are kept apart. The registry states facts, so a `contract` row's claimed consumers must equal the clients that really bind the name — a named import from the package or one of its subpaths, a re-export from it, a member read on a namespace or dynamically imported module object, a destructured dynamic import — and an `internal` name that a client binds must be re-registered as `contract`; a client-side declaration that merely shares the name is not consumption. The retirement check asks only whether removing the name could break a client, so it also counts a plain identifier match in any file importing the package, and, when a client re-exports the whole package, any appearance of the name in that client's sources; the guard prints both counts. The guard prints the ref, commit and commit date of everything it read, and it shares one repository reader with the same-name shadow guard, so both guards judge the same snapshot of a client. A client that is absent or is not a git repository is reported as a skipped section, never as a client with no consumers, and a dangling or unreadable branch ref is a fault rather than a branch quietly left out. The reader proves itself on a throwaway repository: an import on a recent branch that is not checked out must be found and attributed to its branch and line, while one on a branch older than the window, in an uncommitted file, or outside the scanned folders must not count. Names that have already left the surface are kept in a per-version `removed` ledger: they must never reappear in the baseline or the registry, and the ledger's versions must not run ahead of the changelog. A name may come back only through a separate `revived` ledger that names it, the version it left in, the version it returns in (later than the one it left in and not ahead of the changelog), a ticket and a reason, and only if it is really back in the baseline; the `removed` ledger itself is never rewritten, and the documentation freshness check reads the same `revived` ledger so a returned name is no longer treated as retired. The `removed` ledger is also checked against history: compared with the same file at the tag of the expected previous release (the newest released version in the documentation freshness check's frozen ledger), every version it listed must still be there, with every name it listed still under that same version — entries can only be added. Only that tag is used: outside a git checkout, when that tag is not present locally, or when it does not carry the ledger, the check is reported as skipped rather than passed, never run against some other tag. Names cannot leave the public surface unrecorded either: every name in the baseline at that same previous-release tag that is no longer in today's baseline must be listed in the `removed` ledger, under the version of the topmost changelog section or under an earlier version as long as it is not a name that has since returned; the check compares the two sets of names rather than their counts, and it is reported as skipped, never passed, under the same conditions. |
 | `scripts/run-wire-refusal-copy-test.mjs` | Two wire refusals read the same way on every client: a cancel's 409 carries one of two codes with opposite dispositions (`conflict.approval_settled` — someone else already decided, go read the result; `conflict.run_not_running` — nothing changed, send the cancel again), an unrecognised or codeless 409 is reported as such rather than guessed, and the submit-side 429 `usage.window_exhausted` is read as a waitable refusal whose wait is stated only when the engine supplied one. `ControlRouter.cancel` raises a distinct safety code for the retry-directly case. A third family covers the refusals that a deny's *settler note* can draw: a deployment that signs the decisions it accepts but does not sign that note, a body whose decision and note contradict each other, and a word the engine does not recognise. All three are refused **before** the approval is judged, so each sentence states plainly that nothing was decided and the approval is still waiting, and each names a different next step — none of them “send it again unchanged”, which would simply be refused again. Two of the three codes already mean something else in this package: one is shared with the plan-review leg, so the reader refuses to claim it unless the caller states that the note really was sent, and the other is split by the field the engine names, because without that field the same code means a review outcome was rejected for its content. The third sentence deliberately does not say the word was misspelled: upstream mints that same field for at least four different reasons, so the only thing it proves is that the engine would not take the settlement details and named which part — which is what the sentence says, and where it points. The three sentences are pinned verbatim rather than by keyword, because a keyword check passes a sentence that tells the reader to send the same body again unchanged, which is the one next step that is certainly wrong. The field itself is read from the bag the SDK keeps additional response keys in, not off the top of the error, since only the hand-built shapes a test would write carry it there. Both decision legs hand the reading back on their outcome, and a leg that never sent the note claims nothing. When a decision fails because the engine cannot read a stored row, the row the engine names is read (from the error's top level, or from the SDK's bag of additional response keys where the real SDK puts it) and carried into the failure reason on all three decision legs, escaped for display.. A billed request refused because the deployment enforces budgets fail-closed and could not confirm the caller's budget with its quota service (`fleet.lease_unavailable`, 503, engine ≥7.106.0) is recognised by its code alone and gets its own sentence instead of the generic server-failure line: the budget could not be confirmed, the request was not accepted and is not at fault, each known cause adds its own clause, and the wait is stated only when the engine supplied one; the turn-error classifier marks the same case, and the engine's own refusal mapping is replayed through the real SDK to prove every cause reads back The router's `cancel_lost_race` refusal carries the same sentence as the reader for its kind (an approval already settled elsewhere, or a conflict the engine did not explain), and there is only one judge of whether a cancel failure is that 409. |
 | `scripts/run-tool-disclosure-progress-projection-test.mjs` | The two wire arms sdk 9.6.0 adds — `tool_disclosure` (name-only tool census: open-set `policy`, `thresholdPercent` absent ≠ default, `deferred`/`activated` full snapshots) and `tool_progress` (one frame, two beats: Bash ticks carry an output tail with `totalLines`/`totalBytes` that come and go together; other tools carry only `elapsedSeconds`) — project to neutral internal arms plus chrome arms. Required keys missing ⇒ `malformed`; bad optional keys drop only themselves; the sub-flow three-key gate keeps child frames off the leader lane; both arms are `required: false` in the arm table with duties stated (the output tail is untrusted raw and must never be fed back to the model). |
@@ -326,7 +326,7 @@ guard still cross-checks the table by name).
 | `scripts/run-device-executor-management-capability-test.mjs` | The engine's device-lane self-description (`capabilities.deviceExecutor`, with its management bit from engine ≥7.88.0), read by one reader the same four-state way as its sibling capability readers: a lane object is reported as present and carries the management bit as one of three literal states (on, off, or absent — an older engine really ships the lane object without the key, and that is never folded into `false`) plus three optional lane facts (protocol version as a finite number, in-flight cap per device as a finite non-negative number, socket path as a non-empty string), each read on its own so a missing or malformed one only drops that fact; the lane object must carry `enabled: true` as the contract says, and anything else is an unreadable shape rather than an invented disabled-lane state; an absent `deviceExecutor` key is not reported, `deviceExecutor: false` is the lane being absent, presence is judged by own-property not truthiness, every key is read as an own data property (accessors are never run), the reading comes only from the capabilities cache's landed response (there is no separate write port), and the package owns two verdicts — whether the lane is open (`yes` for any lane object, `no` for the lane being absent, otherwise `unknown`) and whether the `/v1/devices` management verbs are usable (`yes` only when the bit is on, `no` when it is off or the lane is absent, otherwise `unknown`) — while the installed SDK's declaration of the lane object is pinned |
 | `scripts/run-run-cancel-context-test.mjs` | The run record's `cancelContext` side-note (engine ≥7.87.3) read structurally, and the cause of a `turn_aborted{engine_error}` classified from machine-readable evidence only: `cancelled` (code `cancelled`, with the cancel-time context when present) / `engine_error` (any other failure code, passed through verbatim) / `run_still_live` (the record is not terminal — a dropped stream is a client-side fact, not the run's cause) / `unknown` (never guessed). An absent `cancelContext` reads as *not reported*, never as "not cancelled"; `elapsedMs` is never folded to 0. |
 | `scripts/run-suspended-reopen-projection-test.mjs` | The durable `suspended` event's `reopened` key read as three distinct states — `reopened` (with the engine's code, verbatim), `not_reopened` (an explicit `null`), `unstated` (key absent or unreadable) — and carried on the HITL bridge's active gate (`currentGateReopen()`), re-read on every `suspended` and cleared with the gate. The stream driver also turns a `suspended` frame that reads `reopened` into the chrome arm `suspended_reopened` (the reading, the single-source reopen sentence when the code is in the reopen family, the frame's event id, and the run id only when the frame itself carries one), pinned end to end: one notice per event id on a stream, emitted in stream order on a replay from the first frame, nothing for a cursor that starts after it, and an unchanged transcript when the host has no chrome sink or the sink fails. |
-| `scripts/run-panel-identity-normalization-test.mjs` | One background subagent has two ids on the wire — the fleet row id tail and the `task_progress` task id (its transcript id). Every panel event goes through one funnel that rewrites the `tick` / `end` task id onto the fleet row's id once a `fleet-row` carrying that `transcriptId` has bound the two ids (the delegating card id is not a key: several agents can share one card), carrying the original as `wireTaskId` and marking `taskIdOrigin`; an unbound tick whose row has not arrived yet waits one beat (bounded) and is released verbatim on the next tick / `end`, when the buffer is full, or after `MAX_HELD_WIRE_TICK_BEATS` other fleet-row / end / sweep events (a `sweep` itself leaves it alone: there is no row to settle yet); a normalized `end` that carries no cycle identity borrows the registering row's, so a late close of a revived task is recognized as stale; records are bounded as a whole and evicted least-recently-hit first, finished ones before running ones (a task that keeps ticking is never evicted by newer registrations); the residency mark lives on the agent's one record, so both ids read and clear the same mark — except that a stale (previous-cycle) terminal never clears the revived agent's mark. Once a tick has been delivered verbatim under its UUID, that UUID is the subagent's key: later fleet rows and fleet-side ends are rewritten onto it (the tail kept in `wireTaskId`), so a consumer sees one row in every arrival order, and a revival — even one with a new transcript id — stays on that key; a late tick from a previous cycle is dropped rather than folded into the revived row, and a fleet row that carries only the delegating card id is its own agent. |
+| `scripts/run-panel-identity-normalization-test.mjs` | One background subagent has two ids on the wire — the fleet row id tail and the `task_progress` task id (its transcript id). Every panel event goes through one funnel that rewrites the `tick` / `end` task id onto the fleet row's id once a `fleet-row` carrying that `transcriptId` has bound the two ids (the delegating card id is not a key: several agents can share one card), carrying the original as `wireTaskId` and marking `taskIdOrigin`; an unbound tick whose row has not arrived yet waits one beat (bounded) and is released verbatim on the next tick / `end`, when the buffer is full, or after `MAX_HELD_WIRE_TICK_BEATS` other fleet-row / end / sweep events (a `sweep` itself leaves it alone: there is no row to settle yet); a normalized `end` that carries no cycle identity borrows the registering row's, so a late close of a revived task is recognized as stale; records are bounded as a whole and evicted least-recently-hit first, finished ones before running ones (a task that keeps ticking is never evicted by newer registrations); the residency mark lives on the agent's one record, so both ids read and clear the same mark — except that a stale (previous-cycle) terminal never clears the revived agent's mark. Once a tick has been delivered verbatim under its UUID, that UUID is the subagent's key: later fleet rows and fleet-side ends are rewritten onto it (the tail kept in `wireTaskId`), so a consumer sees one row in every arrival order, and a revival — even one with a new transcript id — stays on that key; a late tick from a previous cycle is dropped rather than folded into the revived row, and a fleet row that carries only the delegating card id is its own agent. The owner ledger and the subagent content store decide whether two ids are the same subagent only through the reducer identity index, so both ids of one subagent read the same owner run, content and delegating card. Cleaning up follows the same identity: reviving an agent clears the earlier cycle's terminal report under every id of that agent (and never another agent's), the delegating card used for a follow-up is the most recently registered statement across the agent's ids (a late repeat of an older id's tick does not bring the older card back), and clearing an agent's content removes every copy it kept, including content parked under the delegating card, with the byte budget given back at the same time. |
 | `scripts/run-prompt-assembled-projection-test.mjs` | The `prompt_assembled` frame (one prepare's prompt-assembly manifest) projected to an internal arm and then to the additive `prompt_assembled` chrome event — the per-section / per-block **character** counts, the mounted tool names and `totalChars`, each key present only when the engine really sent it (the frame's `constitution` is deliberately not carried: no consumer asks for it today, and every published key is a contract to keep). The manifest carries **no token counts** anywhere upstream, so this projection mints none: a token figure derived from characters would be an invented number, and the engine's own estimate lives on `context_usage.sections[].tokens` (same id wordlist, joinable). Bad rows are dropped one by one, and a face that loses every row reads as an absent key rather than an empty array — so an absent face means only "this event carries no readable view of it" (an absent upstream key, an empty array and a fully filtered list all land on the same shape) and is never reported as a diagnosis about the engine. `blocks[].id` and `sections[].id` are two different wordlists with a many-to-one relation, and the token join against `context_usage.sections[].tokens` only holds when both sides carry a section view. A frame with no readable composition key at all is malformed, ids and slots are read as an open set, one chrome event per frame with zero transcript rows, several prepares per task are all handed over (de-duplication — "take the last one" — is the host's move), and the lane is told honestly (`parentToolCallId` ⇒ subagent lane; a frame attributable only by `sourceTaskId` / `bgAgentId` is not surfaced on the main lane). Both entry points obey the same rule: the adapt layer rebuilds every row too, so a host pipeline (or a replayed transcript) that feeds the raw frame straight into `adapt()` cannot smuggle extra keys (`tokens`, digests, aliases), a negative `chars` or a `null` row into the chrome payload, an empty array does not count as a composition face, the identity keys are snapshotted once on both paths (read exactly once each, a throwing accessor rejects the whole frame — reading one twice is what lets an accessor frame land on a different lane on each path), and the two paths are compared verbatim so the two readers cannot drift. |
 | `scripts/run-compaction-outcome-projection-test.mjs` | The `compaction_outcome` frame (a compaction that did **not** end as compacted: mooted by the task ending, failed, …) projected to an internal arm and then to the additive `compaction_outcome` chrome event — `outcome` required and verbatim (open set), `trigger` / `reason` present only when the engine sent a non-empty string, malformed frames dropped, zero transcript rows, the lane told honestly (`parentToolCallId` ⇒ subagent lane; a frame attributable only by `sourceTaskId` / `bgAgentId` is not surfaced on the main lane). |
 | `scripts/run-approval-card-retract-test.mjs` | The approval card's **decision-free retraction** and the in-stream frame leg's **outcome hand-back**: a host that must withdraw a card that no longer has a decision channel (session switch, engine switch, a tracker reporting the ask gone) answers `{ kind: 'retracted' }` and the package sends nothing on any of the three legs (in-stream frame, suspended ask, durable park); the outcome is the `ApprovalResolution` `{ kind: 'not_sent', cause: 'retracted' }` (`isApprovalDecided` answers false; the pre-0.98.0 reading of `decision: 'unresolved'` plus a `retracted` flag is withdrawn), `aborted` resolves `unsettled` / `interrupted` after exactly one posted deny, `failed` keeps its fail-closed deny, and `onToolApprovalOutcome` hands every in-stream outcome back to the host exactly once, tolerating a throwing or rejecting callback. The former approval-outcome note helper (`approvalOutcomeNoteOf`) is retired with CC-77: read `isApprovalDecided`, `approvalNotSentCause` and `approvalResolutionDetail` instead. |
@@ -375,13 +375,13 @@ guard still cross-checks the table by name).
 | `scripts/run-crash-converged-projection-test.mjs` | The `crashConverged` read face on `GET /v1/approvals` (L-38): what the *previous life* of a crashed local engine left behind, projected for every client. Three judgements are pinned. First, **absence is not an empty list** — a missing key (an older server, deps not present, or a carrier that is not an array at all) returns `undefined`, and the client renders nothing; an empty array returns a present zero-count object, which is the server actually saying "none". Folding the first into `{total:0}` would have the client assert "nothing was left behind" on a surface a person uses to decide whether it is safe to re-run something — the worst possible direction for a false statement — so the two cases are pinned to different **return shapes** and a test asserts the two verdicts are unequal. Second, bucketing is a **four-term conjunction**: `orphanState === 'pending'` *and* `resumeSafe === true` *and* both approval-evidence keys (`originalDecision`, `decidedAtMs`) absent. A fifth term rejects any row carrying an **accessor**, and accessors are never invoked at all — reading one means synchronously running someone else's code, and `catch` catches throwing, not *never returning*, so a looping getter would pin the startup thread forever (the row cap does nothing against that shape). The same rule covers the three untrusted reads outside the row as well — the envelope's `crashConverged` key, the carrier's `length`, and every numeric index are read as own property *descriptors* and only data descriptors are used, so accessors and prototype entries read as absent and are never invoked. Such a key is treated as absent: if it was a required field the row is counted as dropped, if it was optional or additive the row survives without it. That also closes the ordering attack, since spreading runs getters in property order and an earlier one could `delete` the approval evidence before it is ever copied (measured before the fix: such a row reached the resume-safe bucket), and the check therefore moves ahead of the read, onto the property descriptors — from which the snapshot is then built directly, because checking descriptors and *then* spreading is two independent observations of the same row, and a non-throwing proxy can make the two `ownKeys` calls disagree (first showing `originalDecision: 'approve'` so the row reads as plain data, then omitting that configurable key so the snapshot loses the evidence; measured before the fix: the dangerous row reached the resume-safe bucket after exactly two enumerations, and after it, one). Keys are written with `Object.defineProperty` rather than plain assignment, because `'__proto__'` is a legal own enumerable key and `o['__proto__'] = x` does not store a value — it calls the prototype setter, letting a row whose own properties are all plain data (so the accessor gate never fires) inject a prototype whose `sessionId` getter deletes the approval evidence from the snapshot during validation; `defineProperty` fires no setter, so the key survives as ordinary additive data and the snapshot keeps `Object.prototype`. A row that simply arrives with a custom prototype is treated the same way, since the snapshot only enumerates own properties: approval evidence sitting on the prototype would never reach it, and a perfectly ordinary object with no proxy and no accessors could otherwise be called safe to re-run — real bodies come from `JSON.parse` and always carry `Object.prototype`, so nothing genuine trips it). Validation itself runs on a **null-prototype** dictionary and the bucketing verdict is carried out of that same pass rather than re-read from the delivered row, because every property lookup on an ordinary `{}` reaches `Object.prototype`: a polluted `sessionId` getter there would delete the approval evidence from the snapshot mid-validation and send the row to the safe bucket (measured before the fix). The row handed to the client is still an ordinary object — the null prototype is an implementation detail of the check, not of the value) — real JSON bodies are all data properties, so only a middle-layer-synthesised payload ever trips it, and it too lands in the human bucket rather than being dropped. The `decided` arm means the human had already approved and side effects may be half-landed, so it always goes to the human bucket, as does `resumeSafe === false` and — the last two terms — any row whose own fields contradict each other, since `pending` claims nothing ran while that evidence says somebody pressed approve. Deciding "not safe" costs one extra question (recoverable); deciding "safe" wrongly has somebody re-run work that already partly happened (not). A 2x2 truth table pins that exactly one cell is resume-safe, so reading either key alone turns red, and the contradictory rows are routed to the human bucket rather than dropped — they are real orphans, and the ones most worth showing. Third, unreadable rows are **dropped and counted**, never thrown and never passed through: the product is declared as `CrashConvergedRow`, so letting a row missing a required field — or carrying one of the wrong type — past would be a lie at the type level, and the closed literal discriminators (`decision` / `cause` / `orphanState`) decide family membership rather than being an open vocabulary. The measuring stick stops at the **type** floor, though: degenerate-but-well-typed values (`ts: NaN`, an empty `toolName`) are kept, because swallowing a real orphan over a decorative field is the worse direction, and the one deliberate exception is `approvalId`, which must be non-empty to be a row identity at all. `dropped` is kept separate from `total` so unreadable rows never inflate "N approvals were affected"; each row is a **one-shot snapshot** — every own enumerable key is read exactly once, and validation, bucketing and the handed-back value all read that same snapshot, so additive upstream keys survive while a **non-idempotent** getter (one that never throws, just answers differently on a second read) can no longer erase the approval evidence between the check and the bucketing (measured before the fix: such a row landed in the resume-safe bucket while its checked value was `"approve"`). Hostile carriers are counted rather than allowed to reject: **every** touch of the carrier is guarded — envelope property reads, `Array.isArray` itself (it throws on a revoked proxy), the `length` read, each indexed read and each row's property reads — and a traversal that dies halfway returns absence rather than a half-counted total. A row that cannot be read never takes the batch with it: its own shape check is inside its own guard, so one revoked-proxy row costs a `dropped` tick rather than collapsing the whole projection to absence — which a client would have read as "this deployment does not offer the surface". Traversal goes by **numeric index, never the carrier's own iterator protocol**, because `for...of` hands the carrier the question of which rows exist: an array carrying an overridden `Symbol.iterator` can yield nothing (measured before the fix: a real orphan became `{total:0}`, which a client reads as "the server said there are none") or swap a dangerous `decided` row for a safe-looking one (measured: `fake-safe` was returned in place of `real-danger`). Row count is capped at 100000 and the cap is checked **before** the walk: requiring only a non-negative integer `length` does not stop a proxy trap reporting a billion, and this surface runs on the startup / `--resume` path, where a synchronous spin freezes the thread (measured before the cap: twenty million rows took 18.3 seconds and twenty million index reads; a billion does not come back). The honest boundary is stated rather than overclaimed — a proxy can still lie in its `length` or index traps, which is the same thing as a host injecting a lying transport — and the widening of `ApprovalsResourceLike.list()` is proven **additive** by really running tsc over a legacy `{ pending }` mock *and* over the real `AgentClient` path — the projector takes `unknown` precisely because a parameter shaped as "an object with an optional `crashConverged`" is a TypeScript weak type that the installed SDK's own `list()` return shape shares no property with, which only a real-client compile would have caught — with a known-red control so a clean run means the checker spoke Every row in the `needsHuman` bucket also carries a closed-set `needsHumanReason` (`approved_then_interrupted` / `pending_unsafe` / `unstable_row` / `contradictory`) derived from the same single read that bucketed it; `resumeSafe` rows never carry it, a same-named key on the supplied row is overwritten or removed, and the bucketing itself is unchanged. The `cause` discriminant is a closed set of two words — a crash, or an orderly shutdown that cut an unfinished step off — and neither the bucket nor the reason ever reads it: every combination buckets identically under both words, a word outside the set still drops the row, and the row reaches the client with its word unchanged and no sentence attached. An exhaustive switch over `cause` that only knows the older word fails to compile, by design. When server fixtures are present, rows minted by the server's own audit store are projected for both generations: the newer one yields the shutdown rows and an untracked step's pending orphan that is no longer marked safe to re-run, the older one yields neither. The one line a person reads about those leftovers is minted here too, by a single speaker: `crashConvergedNotice(projection)` takes the projection as it is (a host that drops rows it has already disclosed passes the two buckets back with `total` recomputed) and produces one English line — how many approvals were still waiting with the recorded cause codes repeated verbatim, then either one sentence per needs-human reason in a fixed order (`NEEDS_HUMAN_REASON_ORDER`, deliberately not the union's declaration order, which is the bucketing priority) with a fallback sentence last and the "check them by hand" closing said exactly once, or the "based on what was recorded" re-run sentence, then "sema has not re-run anything" and a note for unreadable records. It never re-judges a row (bucket membership is taken from the projection as given), it says nothing for an absent, empty or unreadable projection — including one whose buckets do not add up to its `total`, so a host that filtered rows without recounting gets silence rather than self-contradicting numbers — and a row whose reason cannot be read only falls into the fallback sentence. `crashConvergedNeedsHumanSentence(reason, n, total)` gives the per-reason sentence, or `undefined` for anything outside the four words or a bad count. The order constant is pinned at compile time to be a permutation of the reason union, and the pin is proven to sit on the thing it guards: the guard edits the source in memory and requires the line of the matching pin to fail when a member is removed from either side or duplicated. When the terminal's source tree is present, the same 9432 projections also run through the terminal's own sentence table and assembly, extracted by syntax tree, and must match byte for byte. |
 | `scripts/run-self-orchestration-denial-test.mjs` | The three judgements behind a **denied self-orchestration request** (server 7.57.0), each of which all three clients would otherwise get wrong on their own. First, whether to retry at all is a **conjunction that may not be loosened**: HTTP 501 *and* an `errorCode` that is **exactly** `capability.self_orchestration_required`. That code shares its shape with every other `capability.*` 501, so dispatching on the prefix would drag "some other capability is not wired up" into the retry arm — those requests do not become acceptable once the two keys are gone, so the client would spend a request and then tell the user the wrong reason. Negative controls cover all four directions: a sibling `capability.*` code, a truncated or suffixed variant of the right one, a codeless 501 (it decides nothing, so it decides nothing — no guessing), and the right code under 500 / 400 / 503 or a string `"501"`. The classifier reads structurally rather than by `instanceof` (a host may inject its own transport; across realms or duplicate SDK instances an understandable error would read as unreadable), so a class instance, a bare `{status, errorCode}` literal and an error carrying those fields on its **prototype** all reach the same verdict — and a hostile proxy or a throwing getter yields `null` instead of throwing, because this classifier runs inside a `catch` block where anything it throws escapes the caller's own guard. Second, removing the intent is a **structural** operation, not wording: `selfOrchestration` sits at the top level while `ultracode` sits under `settings` — two different stamping legs — and a client hand-writing `delete` will miss the second one, which costs the user the same failure twice. The single stripper is pinned to touch exactly those two: other `settings` sub-keys and their values survive byte for byte, `deferTools` is left alone (pulling `Workflow` out would be a behaviour change, not a removal of intent), additive unknown keys survive at both levels, the input object is never mutated, `settings` is only dropped entirely when `ultracode` was really there and nothing else remains (an already-empty one is left as is), a non-object `settings` is not touched at all, an `ultracode` that only exists on the prototype does not count, and the whole thing is idempotent. The end-to-end leg runs a real `buildTaskRequest` product through it and asserts the stripped body still passes the registration gate key by key. Third, on the capabilities body, **absence is not "switched off"**: a pre-7.57 server has no `workflowsGate` key at all, so reading absence as "the engine says no" asserts something the server never said, and the mirror-image disease is folding an **unrecognised** `denial` into `null`, which would have the client render "nothing was denied" when the truth is "denied, for a reason I do not recognise". Five shapes are pinned — caps unreadable, gate absent, closed-set member, unknown value, accessor — with the unknown arm carrying the raw token (or an empty one when the value is not even a string) and never collapsing to `null`. All four untrusted reads go through own **data descriptors** only, and the guard pins the getter invocation count at zero, since `catch` catches throwing but not *never returning*; a descriptor trap that throws and a revoked proxy both yield honest absence rather than an exception — though *what* absence means differs by field, and the guard pins that split rather than a blanket rule: an accessor on `workflows`, `workflowsGate` or `engineCan` reads as absent, while an accessor on `denial` reads as `{unknown:''}`, because a key that is **not there** is the gate saying "nothing was denied" whereas a key that is there but cannot be read is "denied, and I could not read why" — folding the second into the first is exactly the false statement this face exists to prevent. Two further pins came out of an adversarial review. The exported retry list is **frozen at runtime**, not merely `as const`: the verdict hands out that same reference, so any consumer splicing it once would poison every later verdict in the process — the guard asserts `Object.isFrozen`, that four different mutation attempts leave it byte-identical, and that a verdict issued *after* those attempts still carries the original two entries. And the classifier reads `denial` only **after** both criteria have passed, since it is not a criterion but an extra field on the verdict: the guard pins the getter invocation count at zero for any error that does not match and at most one for an error that does. The scope line is drawn explicitly rather than overclaimed — "no getter ever runs" holds for `projectWorkflowsGate`, which reads **wire JSON** where every field is an own data property by definition, but not for the classifier, which reads a **thrown value** that may well be an SDK `APIError` class instance carrying `status` and `errorCode` on its prototype; insisting on own data descriptors there would report a perfectly readable error as unreadable, so that side promises only that it never throws. A final pin covers the **integration document's own worked example** rather than the library: the shipped SDK's `tasks.stream()` is an `async` generator, so calling it issues no request at all — the POST happens inside `streamRaw` on the first iteration, and a `try` wrapped around the `stream(...)` call itself can never catch the 501. A client following a submit-shaped recipe on the streaming leg would never run the classifier, and the whole strip-and-retry path would silently do nothing. The guard drives the **real** `TasksResource` against a fake transport, offline, and pins both halves: the synchronous leg is in flight the moment it is called, the streaming leg has issued zero requests after the call and raises on the first `next()` — and it does so through the **real** error path, with `openStream` returning an actual 501 `Response` that the SDK's own `errorFromResponse` turns into the typed error, pinning the `openStream`→`errorFrom` call order so a transport that stops minting `errorCode` cannot pass. The documented recipe is then **executed** rather than keyword-counted: exactly one retry, a second body that really lost both keys while every other setting survives byte for byte, the caller's own request object left untouched, one disclosure and only one, a second 501 propagating with the request count still at two, and — after the first 501 — an abort leaving the count at one with nothing disclosed. A last leg is type-level: `stripSelfOrchestrationIntent` carries an SDK `TaskRequest` overload, because the wide `Record<string, unknown>` form erases the caller's type and the document's "strip and resubmit" line would not compile without an unsafe cast; a real tsc run over a virtual file proves both the narrow and the wide path, with a known-red control — and it compiles the document's two recipes **verbatim**, extracted from the section itself, because a recipe that does not compile is a recipe that was never given: `{ transientOk: true, signal }` is a TS2379 under `exactOptionalPropertyTypes`, which no amount of prose review had caught. The last thing pinned is the one that would have been quietest of all: the SDK's `stream()` returns only on a `done` or `failed` frame, so a stream truncated mid-run — or yielding nothing at all — ends the `for await` just as normally as a completed one. The documented `runOnce` therefore tracks whether it ever saw a terminal frame and raises when it did not, the guard's success fixture emits a real terminal and asserts the handler received it, and a truncated-stream control asserts that shape is reported as a failure with no retry and nothing disclosed. That terminal-frame rule then needed one more turn of its own: the underlying reader returns *normally* when the signal is aborted, so the check as first written rewrote a user's cancellation into a generic stream fault — a client keying off `AbortError` to suppress the error would instead have shown a failure, or resubmitted. Cancellation is therefore checked first, a real-SDK case aborts from inside the handler and asserts the original `AbortError` survives with no retry and nothing disclosed, and the document is checked for that ordering. The harness runs the documented `handle` and `transcript.note` as real spies rather than pushing frames itself, the drive loop rethrows exactly as the document does, and the disclosure ledger is proven to be the caller's own array by a positive identity assertion — without which the cancellation leg's "nothing disclosed" would have been vacuously true. Each recipe is compiled **on its own**, with a preamble that declares only what a host supplies and injects no library symbol, since compiling them together let the second one borrow the first one's imports, and the preamble's own types are decoupled from what the recipes import so the "remove the imports and it must fail" control fails for the right reason — which is checked by attribution, not merely by redness. Ordering is the last thing to get right: the cancellation check must come before the truncation error but **both** must sit behind the terminal-frame test, because a cancellation that lands after the run already reported `done` would otherwise overwrite a real outcome — one that may have already had effects — with "cancelled", and a person reading that will run it again. Aborting from inside `handle(done)` and `handle(failed)` are both pinned to still report success, and the ordering assertion is anchored inside the streaming `runOnce` body rather than the section, since the section's first `throwIfAborted` belongs to the synchronous recipe and would have made a reversed streaming recipe pass — and that ordering check is now anchored on the TypeScript AST rather than on text, since a comment reproducing the two statements in the right order let a genuinely reversed body pass. One more timing fact had to be written into the recipe: a single SSE read buffers several frames and the SDK yields them back to back, so checking the signal only after the loop lets a cancelled run keep consuming the rest of the chunk — measured, an abort inside `handle(turn_start)` still swallowed the `done` that followed and reported success. The recipe therefore re-checks after every non-terminal frame. Finally, the behavioural matrix is no longer run against a copy of the recipe: both recipes are extracted from the document, transpiled, and **executed** with injected host objects, so the disclosure assertion really exercises the document's own `transcript.note(disclose(...))` line, and the synchronous leg gets the same full matrix the streaming one does |
 | `scripts/run-package-hygiene-test.mjs` | Everything `package.json` `files` ships — dist JS/typings and the Markdown docs — is screened line-by-line against a deny-list of strings that must never reach a public tarball (internal hostnames, codenames, person names, collaboration-process words, other repos' ledger ids and repo names; opaque ticket ids `CC-nnn` and post numbers `[nnnn]` are allowed as traceability references). Since 0.77.2 the build strips comments (`removeComments`; enforced by `run-dist-comments-test.mjs`), so what this gate screens in dist is code, string literals and type-level text. Markdown docs are enforced forward-only (CHANGELOG from 0.77.2, the integration doc from §81) because published sections are frozen. From 0.88.0 every shipped file is also scanned in full, frozen sections included, for merge-conflict markers at the start of a line (`<<<<<<< `, a bare `=======`, `>>>>>>> `, and the seven-bar base marker of three-way conflicts). From 0.88.0 a short local list of release-process phrasings (integration-branch and review-round wording) is screened as well, forward-only from the 0.88.0 sections.
-| `scripts/run-integration-doc-freshness-test.mjs` | The **integration contract** (`docs/INTEGRATION-CLIENTS.md`) and the **changelog** (`CHANGELOG.md`) checked against the code, because a document with no guard rots — this one had a whole nest of drift found on it within a day of being written. Five directions, each a claim a machine can actually evaluate. (1) *Counting discipline*: the version-anchor row for the guard count may no longer carry a hand-copied number at all — it changes every time a guard is added, and writing it down is planting a timer; the export counts that are still hand-copied (the surface total, the test-hook count, the sentence describing the surface's internal composition, the sum of the sixteen domain rows, and the three sub-counts) are each compared against a value **derived** from `public-export-baseline.json`, which is the drift a human reviewer caught last time. (2) *Coordinates alive*: every `src/` `scripts/` `docs/` path the doc quotes must be on disk **and tracked by git** — on disk is not in the repo, and a doc that points readers at a file living only in its author's working tree sends every clone to nothing. A file landing in the same commit takes a named carve-out that **stops applying** the moment the file is really tracked (it can no longer let anything through, and the guard prints a line asking for it to be deleted) — deliberately not a red, since turning red on the very commit that lands the file would just manufacture a break that only a follow-up commit could clear. (3) *Arm tables*: the `hitl_out_of_slice` row and the `not_in_slice` fenced list must equal, name for name and in **both** directions, the case labels that really fall into those two buckets — read through the **TypeScript AST**, since which bucket an arm lands in is decided by the argument to `nothing(...)` and by nothing a comment says. The extractor is anchored to the one production projector: exactly one function named `eventToSdkMessage`, exactly one `switch (ev.type)` inside it, and no repeated case label — anything else is a broken anchor rather than a verdict, because a second same-shaped switch elsewhere in the file would otherwise overwrite the real one's conclusions and leave the doc agreeing with a switch nobody runs. The list is delimited by a machine-readable fence rather than by section headings, because the same section also names the terminal arms as a counter-example and prose boundaries cannot tell a member from a foil. (4) *Released sections are frozen*: an **append-only ledger** carries every version ever published — its number, the commit it was published from, and the sha256 of its section — and each one is checked, not just the current release, since pinning only the latest would set every earlier version free the moment the next one ships. The ledger cannot vouch for itself either: each recorded hash is **re-derived from that release commit** through git, so editing an old section and its constant together no longer passes — and the commit the row names is in turn checked against the `gitHead` npm recorded at publish time, which is the one value this repository cannot rewrite, so pointing an old version at a freshly written commit does not pass either. The *set* of versions that must be frozen comes from the registry too, so deleting an old row together with its section — which would otherwise remove that version from every set the guard looks at — is red rather than invisible. A failed registry call is classified rather than swallowed, and the classification consults the registry's own status code *before* it considers connection-level symptoms, so an auth refusal whose body happens to mention the network is still red rather than a skip. The version set is compared as full SemVer including prereleases — matching only `x.y.z` would silently drop a published `0.30.0-beta.1` and reopen the very hole this direction closes — and section headings are matched on a whole-version boundary so a stable release cannot bind itself to the release-candidate section sitting above it. Publishing itself is a two-phase protocol rather than a paradox: before a release, exactly one row may be marked pending and must name the current `package.json` version, exempt from the checks whose inputs do not exist yet; once the registry has that version the row must be promoted, so the temporary state cannot survive its own release. And because the pending exemption rests entirely on "this version is not out yet," it is refused outright when the registry cannot be reached to confirm that — an unverifiable premise is not a licence. Three reverse directions close the rest: a section claiming to be released but absent from the ledger, a ledger entry whose section has vanished, and a `package.json` version that was never frozen. Publishing appends a row; it never rewrites one. (6) *Sentinels*: the readers §5a hands hosts for "is this port installed" are checked against what the source actually declares it returns — `hasXxx()` is a `boolean`, the card port / HITL surface / wire target return `T \| null`, the `installHost` family returns `T \| undefined`. Testing a `null`-returning reader for `!== undefined` is *always true*, and a self-check that passes whether or not the port is installed is worse than none, because hosts retire their own fallback on the strength of it. Both directions are red: an implementation that changes its sentinel without the doc following, and a doc that names the wrong one. The roster covers the zero-argument readers and their `*For` variants alike — a multi-session host reads the variants, so leaving them off would let exactly the surface desktop depends on drift unwatched — and the §5a table and the §8-B checklist line are each checked against the source, because hosts tick the checklist, and a guard that only watches the prose table misses the line people actually follow. (5) *Packaging*: the README ships with the package and opens by pointing hosts at the integration doc, and the checklist names two more files as required reading before an upgrade — all three must really appear in the `npm pack` manifest, or an npm consumer follows a relative link that npmjs rewrites onto a private repository. Missing tooling never takes the whole verdict down with it: when git, npm or the registry is unreachable those legs print the `SKIPPED-SECTION` marker and the rest still judges, while a release commit the ledger names but git cannot resolve is red rather than skipped. The guard says in its own header what it does **not** do: it judges counts, coordinates, arm sets, released bytes and the packing list — whether a sentence is *right* is still for review and for the hosts to report (7) *Retired names*: every name in the per-version `removed` ledger of `scripts/export-liveness.json` may appear in the live sections of the integration doc only where a retirement note follows the name inside the same clause (or the table row's label cell is itself a retirement label); the scan is by identifier boundary after invisible text (HTML comments, link targets, reference-link labels, tag attributes) has been stripped, so a signature line in a code block, an inline `NAME = 4096`, a hidden note, or a note that belongs to a neighbouring name all count as a bare recommendation and go red. Frozen sections (a numbered section whose heading carries a version already recorded as released) are historical and never rewritten, so a retired name there is allowed only if the live retirement catalog has a row for it: a retirement label, the name, the version it left in (matching the ledger) and what to use instead, next to the sentence stating that frozen sections are historical records and the catalog is authoritative. A section whose version cannot be read, or is not yet released, is judged as live. (9) *Known-limits tables*: every row of every table in `docs/KNOWN-LIMITS.md` must split into as many cells as its header — a pipe character inside inline code still splits a Markdown table cell unless it is escaped, which silently shifts the trigger and upgrade columns. Also (0.98.0) every `## §NN` heading carries no repeated topic phrase and the in-flight heading stays under 1200 characters, so a merge that concatenates heading lines is caught. |
+| `scripts/run-integration-doc-freshness-test.mjs` | The **integration contract** (`docs/INTEGRATION-CLIENTS.md`) and the **changelog** (`CHANGELOG.md`) checked against the code, because a document with no guard rots — this one had a whole nest of drift found on it within a day of being written. Five directions, each a claim a machine can actually evaluate. (1) *Counting discipline*: the version-anchor row for the guard count may no longer carry a hand-copied number at all — it changes every time a guard is added, and writing it down is planting a timer; the export counts that are still hand-copied (the surface total, the test-hook count, the sentence describing the surface's internal composition, the sum of the sixteen domain rows, and the three sub-counts) are each compared against a value **derived** from `public-export-baseline.json`, which is the drift a human reviewer caught last time. (2) *Coordinates alive*: every `src/` `scripts/` `docs/` path the doc quotes must be on disk **and tracked by git** — on disk is not in the repo, and a doc that points readers at a file living only in its author's working tree sends every clone to nothing. A file landing in the same commit takes a named carve-out that **stops applying** the moment the file is really tracked (it can no longer let anything through, and the guard prints a line asking for it to be deleted) — deliberately not a red, since turning red on the very commit that lands the file would just manufacture a break that only a follow-up commit could clear. (3) *Arm tables*: the `hitl_out_of_slice` row and the `not_in_slice` fenced list must equal, name for name and in **both** directions, the case labels that really fall into those two buckets — read through the **TypeScript AST**, since which bucket an arm lands in is decided by the argument to `nothing(...)` and by nothing a comment says. The extractor is anchored to the one production projector: exactly one function named `eventToSdkMessage`, exactly one `switch (ev.type)` inside it, and no repeated case label — anything else is a broken anchor rather than a verdict, because a second same-shaped switch elsewhere in the file would otherwise overwrite the real one's conclusions and leave the doc agreeing with a switch nobody runs. The list is delimited by a machine-readable fence rather than by section headings, because the same section also names the terminal arms as a counter-example and prose boundaries cannot tell a member from a foil. (4) *Released sections are frozen*: an **append-only ledger** carries every version ever published — its number, the commit it was published from, and the sha256 of its section — and each one is checked, not just the current release, since pinning only the latest would set every earlier version free the moment the next one ships. The ledger cannot vouch for itself either: each recorded hash is **re-derived from that release commit** through git, so editing an old section and its constant together no longer passes — and the commit the row names is in turn checked against the `gitHead` npm recorded at publish time, which is the one value this repository cannot rewrite, so pointing an old version at a freshly written commit does not pass either. The *set* of versions that must be frozen comes from the registry too, so deleting an old row together with its section — which would otherwise remove that version from every set the guard looks at — is red rather than invisible. A failed registry call is classified rather than swallowed, and the classification consults the registry's own status code *before* it considers connection-level symptoms, so an auth refusal whose body happens to mention the network is still red rather than a skip. The version set is compared as full SemVer including prereleases — matching only `x.y.z` would silently drop a published `0.30.0-beta.1` and reopen the very hole this direction closes — and section headings are matched on a whole-version boundary so a stable release cannot bind itself to the release-candidate section sitting above it. Publishing itself is a two-phase protocol rather than a paradox: before a release, exactly one row may be marked pending and must name the current `package.json` version, exempt from the checks whose inputs do not exist yet; once the registry has that version the row must be promoted, so the temporary state cannot survive its own release. And because the pending exemption rests entirely on "this version is not out yet," it is refused outright when the registry cannot be reached to confirm that — an unverifiable premise is not a licence. Three reverse directions close the rest: a section claiming to be released but absent from the ledger, a ledger entry whose section has vanished, and a `package.json` version that was never frozen. Publishing appends a row; it never rewrites one. (6) *Sentinels*: the readers §5a hands hosts for "is this port installed" are checked against what the source actually declares it returns — `hasXxx()` is a `boolean`, the card port / HITL surface / wire target return `T \| null`, the `installHost` family returns `T \| undefined`. Testing a `null`-returning reader for `!== undefined` is *always true*, and a self-check that passes whether or not the port is installed is worse than none, because hosts retire their own fallback on the strength of it. Both directions are red: an implementation that changes its sentinel without the doc following, and a doc that names the wrong one. The roster covers the zero-argument readers and their `*For` variants alike — a multi-session host reads the variants, so leaving them off would let exactly the surface desktop depends on drift unwatched — and the §5a table and the §8-B checklist line are each checked against the source, because hosts tick the checklist, and a guard that only watches the prose table misses the line people actually follow. (5) *Packaging*: the README ships with the package and opens by pointing hosts at the integration doc, and the checklist names two more files as required reading before an upgrade — all three must really appear in the `npm pack` manifest, or an npm consumer follows a relative link that npmjs rewrites onto a private repository. Missing tooling never takes the whole verdict down with it: when git, npm or the registry is unreachable those legs print the `SKIPPED-SECTION` marker and the rest still judges, while a release commit the ledger names but git cannot resolve is red rather than skipped. The guard says in its own header what it does **not** do: it judges counts, coordinates, arm sets, released bytes and the packing list — whether a sentence is *right* is still for review and for the hosts to report (7) *Retired names*: every name in the per-version `removed` ledger of `scripts/export-liveness.json` may appear in the live sections of the integration doc only where a retirement note follows the name inside the same clause (or the table row's label cell is itself a retirement label); the scan is by identifier boundary after invisible text (HTML comments, link targets, reference-link labels, tag attributes) has been stripped, so a signature line in a code block, an inline `NAME = 4096`, a hidden note, or a note that belongs to a neighbouring name all count as a bare recommendation and go red. Frozen sections (a numbered section whose heading carries a version already recorded as released) are historical and never rewritten, so a retired name there is allowed only if the live retirement catalog has a row for it: a retirement label, the name, the version it left in (matching the ledger) and what to use instead, next to the sentence stating that frozen sections are historical records and the catalog is authoritative. A section whose version cannot be read, or is not yet released, is judged as live. (9) *Known-limits tables*: every row of every table in `docs/KNOWN-LIMITS.md` must split into as many cells as its header — a pipe character inside inline code still splits a Markdown table cell unless it is escaped, which silently shifts the trigger and upgrade columns. Also (0.98.0) every `## §NN` heading carries no repeated topic phrase and the in-flight heading stays under 1200 characters, so a merge that concatenates heading lines is caught. (0.99.0) The per-table length numbers (`ENGINE_NOTICE_CODES: … // N`) now live only in a standing section of the integration contract (`### 2x`, the table-length ledger) and are checked against the source tables there; a length line anywhere else, released sections included, fails the gate, so adding or removing one word edits one live line and never a released section. And (CC-784) in the in-flight changelog section, every `§NN XX-` entry prefix named by a `> 主题:` topic line must have at least one entry line in the same section, so a merge that silently drops entry lines while the topic line survives is caught. |
 | `scripts/run-type-superset-ledger-test.mjs` | The type/wire **superset ledger** (`docs/type-superset.json`): positions this package adds on top of a CC-shaped contract, each carrying the evidence for what CC's own type surface does or does not have there. Completeness is deliberately uneven and the ledger says so. The `_sema_*` private-key class is checked in **both** directions (a key in the source that never entered the ledger is red, naming key and file; a ledger row whose key left the source is red) — but only for keys written as literals, which is the convention the ledger mandates. A key assembled by string arithmetic is beyond what any static rule can enumerate, so the guard fails closed on every shape it *can* decide (a bare `_sema_` prefix is red wherever it appears, save one pinned guard site) and leaves the rest as a convention violation for review to catch, rather than claiming a completeness it does not have. The two hand-surveyed classes are only checked for coordinate and evidence integrity, never discovered. Both directions read the source through the **TypeScript AST**, not a text scan, and they read two different sets out of it. A *key site* is an identifier, or a string whose whole value is the key — so `'_sema_decision-v2'` is carried whole rather than truncated at the first non-identifier character into some *other* key that happens to be registered. A *mention* is the key appearing inside a longer string, which is prose, not usage. The staleness direction counts key sites only: a comment or a doc sentence left behind after the last real mint site is deleted must not keep the row alive (mutation-proven — with both the comment and the prose string untouched, removing the one real site turns the guard red). And because a prefix can be concatenated or interpolated into a key no static set will ever see, the bare `_sema_` literal is refused outright rather than traced: every occurrence is red except the single inline `startsWith` guard the sanitizer needs, because the set of expressions a bare prefix can travel through on its way to a concatenation is open-ended and enumerating it is always one form behind. Every row's `host` must still resolve, with the key being a real **member of that declaration** rather than a string occurring somewhere in the same file — `governanceForced`/`delegation` each live on two different shapes in one file, and a member commented out is a member deleted, which a text-shaped check happily reads as still present. And the direction worth the most: each machine-form `ccAbsenceEvidence` is re-derived from the row's own `key` — the ledger's recorded string must match that derivation verbatim, since a row quietly witnessing `\bnever_present\b` is green forever while watching nothing (mutation-proven: the same edit passes the unbound form and is caught by the bound one) — and the check runs against the names the installed `@sema-agent/agent-types` `.d.ts` set actually declares, parsed with the TypeScript AST rather than grepped, so a name CC merely mentions in a comment cannot force the row into the manual escape hatch and thereby retire the very witness that was supposed to fire the day CC declares that name for real. That escape hatch is gated by an allowlist living **in the guard**, not the ledger, so claiming it costs a reviewed diff. Missing material never reads as a pass, and the verdict splits by *why* it is missing: no TypeScript parser skips the suite before it starts; a missing `agent-types` still runs and prints the first three directions, then exits **1** when `package.json` declares the mirror but it is not installed — a broken install must not retire the repository's only "the day CC declares this name" alarm, and reporting it as a skip would leave "never evaluated" and "evaluated, no drift" indistinguishable to the runner — and exits 3 only when nothing declares the mirror at all, which is the one case where the direction genuinely does not apply. Either way a run that evaluated no witness is never counted as one that did. When the mirror *is* present its **installed version** is witnessed too (the two declared floors must agree with each other and the installed copy must meet them), since four preflight probes are satisfied by an arbitrarily stale mirror — they prove the extractor speaks, not that it is current. Every direction carries a positive control — known-present CC symbols, a comment-only sample proving the extractor distinguishes declaration from mention, and synthetic corpora fed through the **same** discriminator function the real verdict uses, so a verdict quietly rewritten to return nothing takes its own control down with it |
 | `scripts/run-rules-side-test.mjs` | The persisted-permission-rules lane's shared decision half. From 0.92.0 the governance surface reads one capability bit, the rules lane: every server at or above the support floor that advertises it also serves the revoke routes, so the second bit is no longer consulted and a lane-only body shows the surface. Failure classification is by **disposition, not cause**: the two 404s (route missing vs. dead ticket) never share a bucket, a 503 `rule_import_retry` means *the ticket is still alive* (the opposite handling of a dead one), and a stale-cursor 400 drops the cursor and re-lists from the top exactly once — never resuming a stale keyset, never surfacing a partial governance list, and never paging past the hard cap. The persist-ack reader is **merged into** `readToolApprovalRespondAck`: the three-state verdict (`persisted` / `refused` / `unknown`) is derived only from an ack that passed the package's structural narrowing, and a half-shaped object such as `{rulePersisted: true}` with no `delivery` reads as `unknown` — the pre-merge shell read would have said `persisted`, which is precisely the double-ledger drift this file closes, so that case is pinned in reverse. The local-allow-rule skeleton pins all five narrowings (whole-tool, tool-name match, literal anchor with the escaped-star counter-example, bare interpreter prefix consulted only for Bash, and the canonical dangerous-pattern overlay) **with their refusal strings byte-for-byte** — the cli's 128-assertion suite anchors the same strings, so a one-character edit here changes observable behaviour on three clients — and asserts the parse is a pure function of its input, because the same call backs both "render the option" and "resolve the selected value" `listAllPersistedRules` needs only `list` (`RulesListFacade`; the other four methods are known to the parameter type as optional members): a synthetic consumer compiled against the built declarations passes a two-method object, a list-only literal, a `Pick` slice, the named type, the full facade, and the full or two-method facade written as an inline object literal — including arrow functions with untyped parameters — all with zero diagnostics, while a misspelled method name in such a literal is still reported. From 0.94.0 a `request.query_invalid` answer is split by whether that request carried a cursor: with a cursor it is still a stale cursor, without one the listing's query itself was refused (`query-refused`, for example a project scope that is not an absolute path), the server's sentence is kept and the listing is not retried. Since 0.95.0 a caller that cannot say whether its request carried a cursor (no second argument, a non-boolean, or a context whose read throws) gets the generic error with the server's sentence instead of a stale-cursor claim: re-listing from the top is useless for a request that carried no cursor, so naming that cause without evidence is withheld. |
 | `scripts/run-park-decision-layer-test.mjs` | The decision layer behind the "stuck behind a card" family, shared by every client. A pending row that is **not in the queue** is three states, not one: a bounded, interruptible re-probe loop distinguishes *a decidable row*, *not born yet* (no positive evidence that anything settled — an empty queue proves nothing) and *settled elsewhere*, always probes at least once so a zero budget keeps the pre-fix semantics verbatim, cuts a hung read face off at the window rather than only noticing afterwards, and reports the honest failure when the window is spent instead of inventing a decision. The decision-note reader is likewise three-state: an explicit `noteRecorded: false` outranks an echoed note body, absence renders **no line at all**, and untrusted note text is flattened and bounded before it ever reaches a renderer. Row routing anchors on the deciding quantity — a row carrying `gateKind: "human"` with `toolName: "Write"` is a tool gate, because `human` is the engine's *generic* "someone must decide", not a synonym for a question — and the queue scan refuses to surface a row it cannot positively prove belongs to this session. A chain that fails after the row vanished is split by whether a card was ever presented: decided-elsewhere, or not-its-turn-yet. A row-level single-flight makes "at most one card per pending item" structural rather than incidental. The resume three-way card pins the option **order** (the zero-effect choice sits at index 0, because the frame carries no default-focus field and a stray Enter must not attach or cancel), renders only options the wired verbs can honour, collapses every ambiguous answer to zero action, omits the liveness line entirely when the engine gave no evidence, and — when there is no card lane at all — prints three real routes and exits on a dedicated code rather than reporting success |
 | `scripts/run-selfheal-reopen-test.mjs` | The 409 active-run self-heal decision chain: `governanceForced` narrows on strict `true` only; triage prefers the wire's `pendingGate.kind` and falls back to the status table (an off-table kind is never guessed into a card arm — hands-off plus the honest wording); a first-sight card makes zero closed/reopened claims and a host presentation receipt of `presented: false` demotes the outcome to reopen-failed; park-row ownership is a fail-closed positive proof (own-run ledger or session id — unprovable is not owned); the three gate-identity key literals live in exactly one mint (`hitl/gateIdentity.ts`, AST string-token scan); the armed-gate presentation ledger is per-session; and the `plan_review` reopen arm shares the arm arm's card body, three-state verdict and delivery pipe, consuming the presentation history once a decision is delivered. The same chain also carries the `running` three-way card: both plan-family gate kinds route to the plan arm and all four ask-family kinds to the ask arm (an off-table kind still never gets guessed into either); the card is offered only for verbs that can actually be honoured and a missing presenter means zero action rather than a silent cancel; a steer is sent **exactly once** with its three delivery outcomes worded apart (a `queued` receipt is the wire correcting the triage input, so the named park word decides which card gets reopened, and an unrecognised park word drives neither arm), and a steer failure is split into *provably not delivered* (4xx) and *delivery unknown*, because telling a user to resend a non-idempotent instruction that may already have landed is how duplicates get made. After a user-chosen cancel, "the session is free" is asserted only from a whitelist of terminal states — park states hold the claim, an unrecognised state word is not a release, a failed read is *unknown* rather than a release, and only a 404 counts as one — and the honest timeout line quotes how long it really waited. The two "card could not be reopened" rows can carry a host-declared way to keep the conversation, which says the card comes back on resume only if it is still waiting: it is placed before the route that abandons it, never offered for an injected submission, while a decision is still on its way, or once the pending approval has been proven gone (the outcome then carries a flag saying so; the proof only counts before the cleanup card is shown, so a fallback after the card carries no flag unless a fresh read finds the run finished, and a recheck that finds the approval back clears it), the host function is not even called in those cases, and it is treated as unavailable when it throws or returns an empty value; a host can also switch off the engine decide route on the interactive rows while the cancel route stays, and with neither given all four rows are pinned byte-for-byte to the text the previous release produced. When the plan reopen refuses because the user closed that card (`dismissedByUser`), the outcome carries the flag and one package-owned line says so — you closed it, a message brings it back — in the typed, the injected and the queued-steer forms, with no way-to-keep and no engine decide route; the reopen port receives `{ trigger }` mapped from the submission origin (typed ⇒ `user`, injected ⇒ `automatic`, absent ⇒ the old single-argument call). When the reopen refuses because no view is subscribed to question frames for that session (`_sema_noPresentationSurface`), the plan and ask outcomes carry `noPresentationSurface`, and the five lines that talk about the card not coming back — both arms' typed line, the queued-steer typed and injected lines, and the injected not-delivered line — add one package-owned clause saying the session has no view that can show decision cards, byte-identical to the plain refusal otherwise; loose values and the other refusals (answer in flight, user-closed card, plain refusal) carry neither the flag nor the clause. For a host that reopens under its own session key with a presentation receipt, the suite drives the whole sequence end to end: subscribing for that key, reopening, and registering the receipt through the per-session call `registerArmedGateFromQuestionIdFor` gives `reopened: true` with the right first-sight reading, a second reopen reads as seen before, a decisive answer (with no extra bookkeeping from the host) makes the next reopen read as first sight again, and a dismissal does not; registering through the default-session call instead leaves the receipt window to run out with `{ reopened: false }`. Every bit the package reads off a host's reopen verdict is read once, inside the guard: a bit whose getter throws (or a proxy that throws) is treated as absent, so the self-heal still settles — with exactly the outcome it would give if that bit were missing — and the rows never throw, including for an outcome the host built itself; a plain verdict lands in the queued-steer outcome byte for byte, key order and any extra keys the host added included, a key whose getter throws is the only one left out, and an own key literally named `__proto__` is copied as a key rather than changing the copy's prototype; symbol keys are copied too and a verdict with a null prototype stays one (compared key by key, value by value and by prototype, not through JSON); and an outcome whose `reopened` property itself cannot be read is phrased as if it had none. The two reopen-failed outcomes carry `errorCode` only when the reopen gave up because a request failed — the outer wire code of that request (a folded policy code is read through `blockedBy`), taken from a reopen port that threw, from the host verdict's `_sema_errorCode`, or from the stale-park arm's own three requests; a later request that changed what the reopen gave up on (the poll finally answered, the row appeared), a failure with no wire code, a reopen port that failed after the caller had already aborted, and the three judgment refusals (answer in flight, user-closed card, no view) all leave it absent, while a read that only confirms (the row is still pending after a failed decision) keeps it; only a request that failed counts, never a field that cannot be read on a record that did come back; and every row is byte-identical with or without it. The error objects the self-heal leg reads (thrown by the run lookup, the steer and the cancel) are read inside the guard too: a `status` or `message` whose getter throws, or a proxy that throws on any read, is treated as absent, so the self-heal still settles instead of rejecting, and the at-most-once failure class reads such an error as delivery unknown; the same goes for the run record a successful lookup returns — when its `status` cannot be read, the self-heal treats it as a record without a status rather than rejecting — and for a steer receipt, whose two readers answer `null` for a field they cannot read. When the delivery of a steer is missing, unreadable or an unknown word, the injected-submission line says the message was accepted but its delivery was not reported and to watch that reply before deciding whether anything needs sending again, instead of the handed-over wording, and it never says nothing is needed, with or without a follow declaration. The `running` three-way card also gives a fourth way out, *attach*, under the same single rule as the resume card (the run is still running and can be taken over): one shared predicate decides for both cards, only a strict `true` supply flag counts, and a host that does not state the flag gets the request byte-for-byte as before (no `canAttach` key at all, which is not the same as `false`). Choosing it hands the decision to the host without sending any write request (no steer, no cancel), the refused message is reported as not sent, and an `attach` answer on a card that never offered it is zero action. The resume card's two cancel closing rows, confirmed and unconfirmed, are mutually exclusive and come from one private mint. |
 | `scripts/run-terminal-identity-copy-test.mjs` | Terminal-state **identity**, in both lanes where a stop gets a name. A run stopped by this deployment's own governance knobs — the open-set `limits.*` family, `output.invalid`, and the `blocked` contract terminal a ReportBlocked agent produces — is not a provider failure, and labelling it `API Error:` sends the reader to check the network, the key and the quota when the handle is the `--max-turns` they passed themselves. Those terminals now render a neutral row; the reverse direction is guarded just as hard, because asserting "this is *not* an API error" on a code the package does not recognise is the same misfiling pointed the other way — a real `gateway HTTP 502`, a `conflict.session_active_run` and any unknown code all keep the `API Error:` prefix, and the row keeps its `_sema_api_error_message` class flag (spelled `isApiErrorMessage` before 0.83.0) so brief-mode visibility filtering does not silently drop it. The second half is who the rejected submission belonged to: the self-heal copy told every caller "Your message was NOT sent … send it again", which is three separate untruths for a system injection (a plan-review outcome, a cron wake-up, a task notification) — not the user's message, and not re-sendable, since a host queue marks those non-editable and non-recallable. The injected form says so instead, and the one sentence that promises re-delivery is pinned to the single disposition that earns it: `selfHealSubmissionDisposition` is the same function the host consults before putting the item back on its queue, so the promise and the behaviour cannot drift apart, and the arms where no card could be surfaced state plainly that nothing was delivered and nothing will retry. Since 0.72.6 the same gate pins the **follow intent** after a steer (): a message handed to a live run only pays off if someone tails that run's own event stream, so `steerFollowIntent` decides from the delivery word whether to tail now, after the pending decision, or only after a wake — and the "watch that run" sentence ("watch that reply" on the rows about follow-up messages sema sent on its own) turns into a factual "sema is following that run" ("… that reply") **only** when the host declares it attached that tail, so a shell that did not wire it can never claim it did. Since 0.83.0 the rows about follow-up messages sema sent on its own carry no engine-internal words and say what happened per delivery shape, promising a resend or "nothing for you to do" only where the code guarantees it |
-| `scripts/run-additive-key-passthrough-test.mjs` | The one disease shape behind two legs: a **closed whitelist / flattening arm** dropping a fact that is already on the wire, while both sides of the seam look correct. (1) The `task_progress` projection carries a registered **key ledger** — a frame populated with every key the service really projects is pushed through the shipped `eventToSdkMessage`, and the set of wire keys that survive must equal the registered pass-through list **name for name in both directions**, so quietly forwarding one more key is as red as quietly dropping one. `model` (the child run's model id, minted by core as `prepared.model.id` and projected by the server since 7.52.1) is the key this batch adds, with the same conditional the server itself applies: a non-empty string or no key at all — an empty string is neither a model id nor "unknown". The ledger is also checked against the fenced list in `docs/INTEGRATION-CLIENTS.md` §3d, so a doc that still says seven keys while the code forwards eight is red rather than merely stale. (2) The decide-failure arms carry the server's S-02 `currentPending` pointer key from a 409 `approval_stale` refusal onto the outcome the host reads. The reader is structural rather than `instanceof`, because the client is host-injected and the class identity is not this package's to assume; a half triple never mints (half a pointer cannot relocate anything), an empty string is not presence, and `checkpointToken` never transits. Both the allow and the deny leg are driven end to end through the real durable approval path — as is the accept-session leg, where a refusal carrying the pointer key must now re-raise instead of silently re-sending the human's answer for the **old** card as a plain approve (one decide call, pointer preserved), while a legacy 400 still falls back exactly as before — and all three flattening points must call the one shared reader — the same-shape residue check that makes "fixed one arm and left the twin" red instead of invisible. (3) The same disease growing on the REQUEST side: the `.mcp.json` → server-spec projection rebuilds each server key by key, and the settings schema deliberately leaves some keys parse-transparent — whatever JSON the file carries reaches the engine untouched, because validating them where the whole domain parses all-or-nothing would let one bad declaration take every server down silently. The whitelist had no row for the newest of them, so an operator's per-tool declarations — the ones the write fence reads — were stripped at the package boundary while both sides looked correct. The criterion is not "is that key handled" but the transparent-key table read out of the INSTALLED schema at runtime, reconciled name-for-name against this leg's ledger, so the day upstream adds a third one this turns red and forces an explicit decision. Behaviour is pinned on both transports, by object identity rather than deep equality (a rebuild would be a second judge), and malformed values must transit UNCHANGED rather than be refused here — the engine refuses them loudly and names the server, whereas a package-side judge can only swallow a declared protection quietly. Absence still mints no key, unknown keys still never reach the wire (the fix is the dropped key, not the gate), and the one transparent key this leg deliberately does not forward is a ledger entry with its own exit condition: it belongs to the deployment plane, and the day the request-plane type declares it the entry's premise is gone and the gate says so The prose probe reason on approval asks (`probeReason`: live frames from server 7.115.0, and the durable pending row's risk descriptor) lands on the same card field from both legs, byte for byte, never as an empty string, read from own keys only. A display field that throws when read (a getter or a proxy trap) is treated as absent: the card is still shown and the decision is still sent. The remember-refused flag (`rememberRefused`, the server saying an allow-for-this-session answer will not be remembered on this card) reaches the same card field from the live frame and from the pending-list row, only as an own strictly-true key; absent, malformed, inherited or throwing values leave it off without blocking the card. The single reader of the card’s remember-refused declaration is a root export — an own key holding exactly `true`, never throwing on hostile input, agreeing with both card legs over the same pool — and no private copy remains. Since 0.96.0 the retry-status leg also carries the provider condition word and its evidence source verbatim to the top of the retry status; no quota reading is derived from them. Since 0.96.0 the configuration-key ledger also accounts for the three transparent connector keys of the 12.2 settings contract: none of them is forwarded on either leg, two because the request shape has no such key, and the per-tool axis override because its meaning on a centrally configured entry is not yet defined by its consumer; each entry carries the condition that would end it. Every withheld key is probed on both transports and with nine value shapes on both legs, the request shape is read with the type checker, and the per-tool axis override's end condition is read by running the current server package's own central projection. |
+| `scripts/run-additive-key-passthrough-test.mjs` | The one disease shape behind two legs: a **closed whitelist / flattening arm** dropping a fact that is already on the wire, while both sides of the seam look correct. (1) The `task_progress` projection carries a registered **key ledger** — a frame populated with every key the service really projects is pushed through the shipped `eventToSdkMessage`, and the set of wire keys that survive must equal the registered pass-through list **name for name in both directions**, so quietly forwarding one more key is as red as quietly dropping one. `model` (the child run's model id, minted by core as `prepared.model.id` and projected by the server since 7.52.1) is the key this batch adds, with the same conditional the server itself applies: a non-empty string or no key at all — an empty string is neither a model id nor "unknown". The ledger is also checked against the fenced list in `docs/INTEGRATION-CLIENTS.md` §3d, so a doc that still says seven keys while the code forwards eight is red rather than merely stale. (2) The decide-failure arms carry the server's S-02 `currentPending` pointer key from a 409 `approval_stale` refusal onto the outcome the host reads. The reader is structural rather than `instanceof`, because the client is host-injected and the class identity is not this package's to assume; a half triple never mints (half a pointer cannot relocate anything), an empty string is not presence, and `checkpointToken` never transits. Both the allow and the deny leg are driven end to end through the real durable approval path — as is the accept-session leg, where a refusal carrying the pointer key must now re-raise instead of silently re-sending the human's answer for the **old** card as a plain approve (one decide call, pointer preserved), while a legacy 400 still falls back exactly as before — and all three flattening points must call the one shared reader — the same-shape residue check that makes "fixed one arm and left the twin" red instead of invisible. (3) The same disease growing on the REQUEST side: the `.mcp.json` → server-spec projection rebuilds each server key by key, and the settings schema deliberately leaves some keys parse-transparent — whatever JSON the file carries reaches the engine untouched, because validating them where the whole domain parses all-or-nothing would let one bad declaration take every server down silently. The whitelist had no row for the newest of them, so an operator's per-tool declarations — the ones the write fence reads — were stripped at the package boundary while both sides looked correct. The criterion is not "is that key handled" but the transparent-key table read out of the INSTALLED schema at runtime, reconciled name-for-name against this leg's ledger, so the day upstream adds a third one this turns red and forces an explicit decision. Behaviour is pinned on both transports, by object identity rather than deep equality (a rebuild would be a second judge), and malformed values must transit UNCHANGED rather than be refused here — the engine refuses them loudly and names the server, whereas a package-side judge can only swallow a declared protection quietly. Absence still mints no key, unknown keys still never reach the wire (the fix is the dropped key, not the gate), and the one transparent key this leg deliberately does not forward is a ledger entry with its own exit condition: it belongs to the deployment plane, and the day the request-plane type declares it the entry's premise is gone and the gate says so The prose probe reason on approval asks (`probeReason`: live frames from server 7.115.0, and the durable pending row's risk descriptor) lands on the same card field from both legs, byte for byte, never as an empty string, read from own keys only. A display field that throws when read (a getter or a proxy trap) is treated as absent: the card is still shown and the decision is still sent. The remember-refused flag (`rememberRefused`, the server saying an allow-for-this-session answer will not be remembered on this card) reaches the same card field from the live frame and from the pending-list row, only as an own strictly-true key; absent, malformed, inherited or throwing values leave it off without blocking the card. The single reader of the card’s remember-refused declaration is a root export — an own key holding exactly `true`, never throwing on hostile input, agreeing with both card legs over the same pool — and no private copy remains. Since 0.96.0 the retry-status leg also carries the provider condition word and its evidence source verbatim to the top of the retry status; no quota reading is derived from them. Since 0.96.0 the configuration-key ledger also accounts for the three transparent connector keys of the 12.2 settings contract: none of them is forwarded on either leg, two because the request shape has no such key, and the per-tool axis override because its meaning on a centrally configured entry is not yet defined by its consumer; each entry carries the condition that would end it. Every withheld key is probed on both transports and with nine value shapes on both legs, the request shape is read with the type checker, and the per-tool axis override's end condition is read by running the current server package's own central projection. Since 0.100.0 the check that an explicitly empty tool allow-list reaches the engine as written is read per server generation (the older generation's accept-list constant and the newer generation's reader assignment are each recognised only in their own form) and its behaviour is run for real on both generations: the server's own request reader keeps the empty list, also when the package's output is fed to it, and the same generation's engine, mounting a stub tool server, mounts both tools when the list is absent, one when it names one, and none when it is empty. The progress frame now forwards its cycle number. |
 | `scripts/run-esc-halt-plan-test.mjs` | The Esc stop decision every client shares: fire the **turn-level** halt first, and escalate to a **run-level** cancel in exactly two cases — the engine itself answered with a 409 from the closed code set (it is saying "there is no in-flight turn here; use cancel for a run-level stop"), or that shot came back with no verdict at all *and* the shell can independently prove a permission card was on screen. Everything else does not escalate. The asymmetry is the whole point and every negative control guards the same direction — deciding *not* to escalate costs the user one more choice on a busy-session card (recoverable), deciding to escalate wrongly tears down a run that was alive and takes every in-flight tool with it (not). So: the closed code set is a **frozen** value, not a `ReadonlySet` — type-level immutability does not stop a consumer's `.add()`, and the guard proves it by really trying to mutate the exported value and then checking the verdict did not drift; the escalation gate is the **conjunction** of that closed set and the 409 status, since honouring the code alone lets a 500 that merely quotes it drive a destructive call; `interrupt.not_held` and `steering.not_running` are deliberately outside the set (the first means *this replica* has no live face — the run may be perfectly alive on another); an unreadable code falls to the no-escalation side; a `parked` flag never overrides a verdict the engine did give, and only strict `true` counts when it did not. The first shot is unconditional by construction — it does not consult `parked`, because the 409 it earns is exactly the verdict the gate wants — and the verdict itself is a closed machine-readable reason word, not display copy. A third escalating case was added once tearing the stream stopped reaping the run: with detach armed, a shot that never lands leaves the run going all the way to the end of the turn, so the Esc the user pressed has no effect at all and nothing on screen says so — the old behaviour had a silent backstop (tearing the stream ended the run) and that backstop is gone. The new fact is held to the same three disciplines as `parked`: it is read only where the engine gave no verdict, it is judged **after** `parked` so an existing host's reason word does not change under it, and only strict `true` counts. Absence is proven to be a no-op rather than asserted — the guard carries its own reference implementation of the previous version's table, runs the full grid through both, requires zero divergence when the new field is omitted, and first shows the comparison really does report a difference on the one cell where the two versions are meant to differ |
 | `scripts/run-peer-frame-projection-test.mjs` | The three engine-injected lanes design/385 puts on the **one** `task_notification` carrier, which are not the same kind of thing at all: a delegated child's uplink (`agentMessage`), another session's message drained from this session's own box (`crossSessionMessage`), and a receipt about one of *this* session's own outbound messages (`crossSessionNotice`). The engine renders none of them inside a `<task-notification>` shell, so a client that projects them as the generic completion card shows "background task finished" while the model read a colleague's sentence — two faces describing different events. The discriminator is pinned to the **typed carrier being present**, never to the `summary` text: those carriers can only be minted by the engine's injection legs (the external `notify()` input is a strict subset of the payload and can wear none of them), while `summary` is filled by every notification there is — so anchoring on text would let any background task impersonate a colleague's message by writing `<agent-message from="…">` into its own summary, and a positive control asserts exactly that payload still projects as the generic card. Fail-closed has two tiers rather than one: a broken **required** field (empty `from`, a non-string `body`, a notice `kind` outside the closed set) returns absence so the caller falls back to the generic card — an honest downgrade where the user still sees the notification — while a broken **optional** field drops only itself, because losing an attribution note and losing a colleague's whole message are not the same magnitude. The provenance side record is **required and must agree on four points** (`kind` matches the lane; `from`/`taskId`/`seq` are present and equal the carrier/payload — each equality is anchored on a core mint site and pinned by the cli wire-anchor A-K24), so a carrier signed with a trusted name but a disagreeing provenance falls back to the generic card; peer bodies pass the same authority-envelope neutralization core applies (`<task-notification>` etc. are defused) so a colleague's text can never seed the resume dedup ledger. Lane precedence copies the engine renderer's own order, because the model already read the frame in that order and a client ordering of its own would put a card on screen that disagrees with the frame the model saw. Rendering and parsing of the transcript line live in the same module and are round-tripped in both directions, including a body carrying a forged closing tag (a parser fooled there hands half a message to the next row) and a quote inside the sender label (which must not forge a second attribute); the notice lane is deliberately kept **out** of the parser, since recognising it would mean anchoring the `[Cross-session …]` prefix and a user typing that same line would be rendered as engine speech. Hostile carriers are read as own **data** descriptors only and accessors are never invoked at all — `catch` catches throwing, not never returning — proven by a counting getter that must stay at zero calls, alongside a revoked proxy and a prototype-only carrier; and four legacy payload shapes assert the no-carrier path is byte-identical to before, which is the executable form of "zero difference for an older host". A re-supplied cross-session message — same task id, status and sequence as the first delivery, handed to the model again after compaction — is not rendered a second time, because the first delivery is still on the user's screen; a record with the next sequence number still renders |
 | `scripts/run-wiring-manifest-projection-test.mjs` | The two end-user facts carried on the engine's `wiring_manifest` frame (`modelGate`: which tools this run's model gate removed and the verbatim restore hint; `autoMode`: whether auto mode is actually armed and the engine's own reason word). Projection: both sections ride as `_sema_`-prefixed superset keys, verbatim, and no SDK-named key is minted; a frame where neither section is well-formed projects to `none/not_in_slice` (no empty arm); `modelGate` needs all three keys and treats `removed: []` as a bad value rather than a reading; `autoMode` needs a boolean plus a non-empty reason that agrees with it, and the reason word is never mapped onto the capabilities vocabulary; the frame is flat (a nested `manifest:{}` wrapper is not a supply); `eventId` rides like every other arm. Adapter: exactly one chrome event on the main lane, a sub-flow frame (any `parentToolCallId`, `null` included) yields nothing, and an absent `eventId` leaves the key absent. Added at receiving time because the shell-side gate could not see this package's behaviour: two mutations (empty `removed` accepted, sub-flow gate removed) had passed the package suite untouched 0.71.0 adds sections F–I: the fourth/fifth/sixth manifest sections (`tools` via the roster reader, `hooks[]` rows dropped one by one when malformed, `lsp` absent unless `mounted` is a boolean), the `tool_roster_delta` arm (narrowed `delta`, `malformed` when `fromDigest`/`roster` cannot be read, host applies it against its own digest), the `context_usage` arm (finite-gated scalars plus `sections[]` rows dropped one by one), and the `WiringManifestMcpEntryView` rename with `MAX_AGENT_SKILLS` gone from the surface From 0.84.0 the `hands` section (whether this leg was assembled with the engine's built-in file and shell tools) is projected as well: a strict boolean `mounted` plus the engine's reason word passed through as written, absence kept as "not reported" rather than folded to either answer; a three-state reader and one sentence source share their opening and closing with the roster-derived sentence, whose bytes do not change. The sentence source answers the not-reported sentence for anything it does not recognise, including a roster-derived reading, and never throws. From 0.88.0 the `fileHistory` section (whether this leg records file history; `on` / `disabled` today, passed through as written for unknown words) is projected as well: present only as a non-empty own string, absence kept as "not reported" rather than folded to `disabled`, malformed values dropped like the server drops them, the same view on the non-streaming submit receipt, and a leg built and published by an installed server package checked end to end when one is available. A `fileHistory` value that throws when read leaves only that section absent; the other sections of the same frame are still delivered on both the streaming and the submit-receipt paths. The server-minted `instructions` section (which instruction file the leg loaded) is projected verbatim as a frozen copy, an explicit empty selection stays an empty list, absence stays absent, a malformed section is dropped without touching its siblings, and a frame carrying only this section still becomes an event. The `fleet.subagentTranscripts` statement (from server 7.115.0: how long delegated sub-agent conversations survive) is passed through verbatim as an open word in a frozen `fleet` section; absence stays absent and is never read as `none`, and a malformed value drops only that section. |
@@ -404,9 +404,9 @@ guard still cross-checks the table by name).
 | `scripts/run-cost-reconcile-projection-test.mjs` | The **end-of-run cost reconciliation** reaching consumers at all. The engine splits a run's spend on the wire — the task's own cost, which deliberately excludes delegated sub-agents, the delegated total itself, and the within-task compaction subtotal that sits inside the own figure — and states two reconciliation identities for them. The package used to project none of it, so a cost view could only ever see one number and under-reported both delegated and compaction spend. Both structures are now projected onto the result as superset fields in the wire's integer micro-currency unit, read key by key, with unreadable keys dropped individually, an entirely unreadable structure omitted rather than emitted empty, and unknown categories passed through since the vocabulary belongs upstream. The delegated cost stays **absent when it was never priced**, never a fabricated zero. The same reader also feeds a terminal chrome arm carrying the three parts plus the reconciled total, so the two faces can never compute different answers; the reconciled total is minted only when both sides are known, and otherwise a discriminator bit says which side is unknown. **The reference field for total cost keeps its meaning** — it remains the task's own spend and the delegated total is not folded into it — because that is a shape the wider ecosystem reads; the reconciled figure is offered beside it, not in place of it. A frame that carries no stats emits no arm at all, and the existing rule that in-stream per-turn usage is not published for sub-flows is pinned unchanged, since delegated spend arrives once, at the end. The bit that says those figures are a lower bound is **per stream**, not per context: the emit context belongs to the caller and may be reused across streams, so a gap observed on one run is no evidence at all about the next one — the observation is held for the duration of one stream and handed to both projection faces by value, and the guard drives a reused context both sequentially and concurrently to prove neither direction leaks |
 | `scripts/run-task-progress-terminal-projection-test.mjs` | The one tick that says a delegated child **finished**. The engine fires exactly one final beat carrying a terminal face, and says in the same breath why it exists — so a consumer sees the row finish instead of watching it vanish after the last running beat — but the package's projection whitelist had no seat for that field and its adapter still carried the older premise in a comment, so the terminal beat arrived byte-identical to another running one: the panel row stayed up waiting for a defensive sweep (which only ever settles rows bound to a card still open this turn) or for a separate notification frame. The status now rides through as an **open set** with the vocabulary left upstream, while the question *which words are terminal* is answered by a closed pair on the adapter side — an unrecognised new word takes the running path, because guessing it terminal ends a row that is still working whereas one extra running beat merely renders late. A terminal beat settles the row directly under the lane proof its binding gives it (not the main lane a notification would use, and not by card id, since the engine is naming a child rather than closing a card), freezes the inline group-row twin in the same beat so a later sweep cannot reset the real tool count, clears the session-resident ledger, and fires the stop hook only for a child whose start really fired. It does not mark the row live or emit a second progress beat, and it shares the settled-row ledger with the other two settle legs so a replay or a double-delivery cannot produce a second end. Three things are pinned **unchanged**: a running beat, an absent status (older engines never send the field, and reading absence as terminal would make every child row disappear on its first beat), and the workflow lane gate, which still runs before any of this |
 | `scripts/run-assistant-arm-identity-test.mjs` | The identity keys on an assistant row, and an explicit account of the two that are **deliberately not** there. What the renderer received was a bare role-and-content object, so a dozen consumer sites downstream were each estimating what the message envelope should have told them. The id is taken from the engine's own event id rather than minted locally, because it has to be **the same value** on the live leg and on a durable replay — a freshly minted one would make a replayed message look new to a host's dedup and to rewind — and when the wire carries none the key is simply absent rather than filled with a random stand-in wearing an identity it does not have; it is also kept distinct from the envelope's own local render key, which is a different identity. The model name comes from what the host pinned when it opened the stream (the request was the host's to build) and is never guessed, since a wrong model name is worse than none once a billing or capability face looks it up. Usage and stop reason are **not** minted on this arm, and the reason is frame order rather than effort: content arms arrive before the turn's closing frame, so at the moment the arm is emitted the engine has not yet said what the round cost — anything put there would be an estimate, which is the very thing this work exists to remove — and synthesising a follow-up assistant update when the real figure lands is also refused, because that shape does not exist upstream and would place a message in the transcript the engine never sent. Their real values leave through the turn's own neutral arm as two superset keys, the usage one reusing the **same single mint point** the footer rollup already folds so the two faces cannot diverge, and the stop reason passed through verbatim as an open set — the machine signal for *was this turn cut short*, previously blind on both the stream and the trace. The existing behaviours beside them are pinned too: no arm at all when usage is wholly absent, and the sub-flow cut-out that keeps a child's turn from driving the leader's face |
-| `scripts/run-text-segment-authority-test.mjs` | The **authoritative segment replacement** on `text_end` (server >=7.75.3). `text_end.content` now goes through the same redactor as `result` and the ledger while `text_delta` stays verbatim, so the two **may differ** — an answer that quoted a credential used to be committed to the local transcript in its unredacted form, because the arm only forwarded the boundary signal. Six timing shapes are pinned, two of which an adversarial review reproduced against the installed engine's real bytes and which the first design got wrong in both directions: a second boundary in the same turn (the per-block case on one provider lane) used to make the first segment's prose vanish, and a boundary that arrives *after* the tool card (the other lane emits it at finalize) used to be read as "this package never handled that segment" and reported nothing at all. Three additive keys, all never-false; the two shapes that look alike are told apart by the second one, because the host's action in them is the opposite. The end-to-end legs drive the real pipeline without hand-inserting a segment commit — doing so is exactly what hid the first defect. A second review round then found two combination timings on top of the first fix — a tool card followed by *more* deltas in the same segment, and a byte count that had been documented as a message count — and both are pinned here too. A third round caught a length that the prose called bytes while the code returned UTF-16 units — harmless in ASCII, and on CJK text enough to leave the credential on screen — plus a backfill ledger that had to be kept in step, so the terminal frame does not re-render the segment a second time — kept in step only where the whole stretch sits in one message, because those ledgers are per-message and a fourth round showed that writing across them charges one message's prose to another. A fifth round settled the whole class into one invariant the guard now checks against the previous release's behaviour: this package only rewrites bytes it is still holding in the current message — once a segment has crossed a package-side boundary it emits the three keys and changes nothing else **0.68.2 (CC-01):** the segment identity is now minted here, not by the host: every committed assistant text row carries a top-level `_sema_segment_id` (stamped once at the `adapt()` exit, so the durable whole-message leg and the streamed-segment leg are covered alike; thinking blocks, tool_use-tailed rows and chrome events are left byte-for-byte), `text_segment_end` carries the same value as `segmentId` before rotating, subagent boundaries never rotate, and a replayed stream yields the same identities. Three mutations (no rotation / no stamping / stamping tool_use rows) each turn the guard red **0.69.0 (CC-02):** the same authority replacement now covers the reasoning face (`reasoning_end`, server >=7.77.0): a thinking block still buffered is swapped whole and its live tail recomputed; one already committed at a boundary (the usual timing, since the first text delta commits it) is left untouched and the host is told the row to replace by its uuid, never re-emitted. Subagent boundaries are ignored and the text-segment identity does not rotate **0.69.1 (CC-09):** the run-stream replay guard still drops a frame whose event id was already seen, but it now reports the drop through the host's dropped-frame sink as `duplicate_seq` instead of vanishing silently (server 7.77.0 reuses the first reasoning delta's id for `reasoning_end`, so that authoritative segment is lost on the print lane until 7.78.1); the interactive adapter has no such guard and keeps receiving it **0.69.1 (CC-10/CC-11):** subagent segment-end frames are fenced on all three identity keys (a frame carrying only `sourceTaskId` no longer masquerades as the leader's), and a reasoning segment that spans tool cards now hands the host every committed row it covers (`committedUuids`) so nothing unredacted is left behind. **0.96.0 (CC-645):** the tail of a segment that the authoritative text had not yet streamed is now emitted *before* the segment-end event, so nothing for the same segment follows it; a host that replays the sequence (append deltas, replace with `content` when `diverged`) shows the segment exactly once, on both the text and the reasoning channel |
+| `scripts/run-text-segment-authority-test.mjs` | The **authoritative segment replacement** on `text_end` (server >=7.75.3). `text_end.content` now goes through the same redactor as `result` and the ledger while `text_delta` stays verbatim, so the two **may differ** — an answer that quoted a credential used to be committed to the local transcript in its unredacted form, because the arm only forwarded the boundary signal. Six timing shapes are pinned, two of which an adversarial review reproduced against the installed engine's real bytes and which the first design got wrong in both directions: a second boundary in the same turn (the per-block case on one provider lane) used to make the first segment's prose vanish, and a boundary that arrives *after* the tool card (the other lane emits it at finalize) used to be read as "this package never handled that segment" and reported nothing at all. Three additive keys, all never-false; the two shapes that look alike are told apart by the second one, because the host's action in them is the opposite. The end-to-end legs drive the real pipeline without hand-inserting a segment commit — doing so is exactly what hid the first defect. A second review round then found two combination timings on top of the first fix — a tool card followed by *more* deltas in the same segment, and a byte count that had been documented as a message count — and both are pinned here too. A third round caught a length that the prose called bytes while the code returned UTF-16 units — harmless in ASCII, and on CJK text enough to leave the credential on screen — plus a backfill ledger that had to be kept in step, so the terminal frame does not re-render the segment a second time — kept in step only where the whole stretch sits in one message, because those ledgers are per-message and a fourth round showed that writing across them charges one message's prose to another. A fifth round settled the whole class into one invariant the guard now checks against the previous release's behaviour: this package only rewrites bytes it is still holding in the current message — once a segment has crossed a package-side boundary it emits the three keys and changes nothing else **0.68.2 (CC-01):** the segment identity is now minted here, not by the host: every committed assistant text row carries a top-level `_sema_segment_id` (stamped once at the `adapt()` exit, so the durable whole-message leg and the streamed-segment leg are covered alike; thinking blocks, tool_use-tailed rows and chrome events are left byte-for-byte), `text_segment_end` carries the same value as `segmentId` before rotating, subagent boundaries never rotate, and a replayed stream yields the same identities. Three mutations (no rotation / no stamping / stamping tool_use rows) each turn the guard red **0.69.0 (CC-02):** the same authority replacement now covers the reasoning face (`reasoning_end`, server >=7.77.0): a thinking block still buffered is swapped whole and its live tail recomputed; one already committed at a boundary (the usual timing, since the first text delta commits it) is left untouched and the host is told the row to replace by its uuid, never re-emitted. Subagent boundaries are ignored and the text-segment identity does not rotate **0.69.1 (CC-09):** the run-stream replay guard still drops a frame whose event id was already seen, but it now reports the drop through the host's dropped-frame sink as `duplicate_seq` instead of vanishing silently (server 7.77.0 reuses the first reasoning delta's id for `reasoning_end`, so that authoritative segment is lost on the print lane until 7.78.1); the interactive adapter has no such guard and keeps receiving it **0.69.1 (CC-10/CC-11):** subagent segment-end frames are fenced on all three identity keys (a frame carrying only `sourceTaskId` no longer masquerades as the leader's), and a reasoning segment that spans tool cards now hands the host every committed row it covers (`committedUuids`) so nothing unredacted is left behind. **0.96.0 (CC-645):** the tail of a segment that the authoritative text had not yet streamed is now emitted *before* the segment-end event, so nothing for the same segment follows it; a host that replays the sequence (append deltas, replace with `content` when `diverged`) shows the segment exactly once, on both the text and the reasoning channel **0.100.0:** one committed transcript row now holds exactly one engine segment. The segment's remaining bytes are committed as its own row right after its segment-end event (no more carrying a finished segment into the next segment's row), every row of a segment carries the same `_sema_segment_id` as that segment's end event, the identity is derived from the segment's first text frame (so a projection that joins at the next segment computes the same value), and a model-round end starts a new identity. A segment that genuinely crosses a tool card keeps one identity across its rows. |
 | `scripts/run-gate-negative-controls-test.mjs` | Whether the registry-shaped guards among the suites above actually turn red when the material they check really breaks — a census found the ones clean enough to rehearse safely (closed sets, mirrors, baselines, floors, a type-shape ratchet) without touching any judgement code. Each is exercised by tampering a disk copy of the real material, spawning the guard's own unmodified script, asserting it exits non-zero and names the disease, then restoring the file byte-for-byte. The guards of the same shape that are too costly to rehearse, and the behaviour/projection suites that red on their own assertions, are catalogued rather than rehearsed — see `docs/GATE-NEGATIVE-CONTROLS.md` for the three tables (all generated from `gates-manifest.json`, which carries each suite's classification), the reasons, and a one-minute manual replay recipe for each blind one. The suite reconciles its case list against that classification by name in both directions, so a case quietly dropped from the array without the manifest following is itself an undeclared blind guard. The backup that makes the restore possible is taken by **exclusive create**: checking for it and then copying are otherwise two steps, and two instances can pass the check together — the later one overwrites the only clean copy with material the earlier one has already tampered, and the rehearsal that promises to leave no trace leaves a permanently corrupted file instead. That interleaving is rehearsed too, in a throwaway directory of its own. A run that is interrupted — by the suite runner's time limit, by Ctrl-C or by its terminal closing — restores what it was in the middle of tampering with before it exits: the rehearsed guard's whole process group is stopped first, so nothing that guard started can write the tampered state back afterwards; the file is restored byte-for-byte from its backup and the backup removed only once the bytes match; the build output is regenerated from the restored sources within the grace period the runner allows (a rebuild that would overrun it is cut short, its compiler stopped, and the run says the build output was not regenerated); and the run exits with 128 plus the signal number rather than passing. Before the first rehearsal the suite also looks for backups left behind by an earlier run that could not clean up — in the trees its targets live in, following a symlinked dependency directory — and if it finds one it refuses to start, names the file and how to check and restore it, and leaves the backup untouched rather than overwriting what may be the only clean copy. |
-| `scripts/run-engine-cap-reader-factory-test.mjs` | The one shared implementation behind every capability reader's ports, exercised as a table. Since 0.95.0 a reading is a lazily computed, frozen projection of the capabilities cache's landed response for that base URL, memoised per landing; a reader has a single read port — the probe tee port, the per-reader test reset and the per-reader `forget` port are all gone, invalidation is `invalidateEngineCaps` alone, and a test that wants a clean slate resets the cache. Every reader in the table runs the *same* criteria (the table length is the source of truth, and a roster check fails the gate if any source file calls the factory without having a row): the four states (nothing landed, present, key absent so not reported, malformed or unreadable once a response has landed, so a distinct unreadable state rather than unobserved); a throwing projection or descriptor read treated exactly like an unreadable one and never escaping; a projection that changes the generation mid-flight answering unobserved without being remembered; a getter at the top level or at any nested position never being executed and reading as malformed; a freshly minted "unobserved" reading on every miss (two misses are never the same object, so a consumer that mutates one cannot taint another base URL); one independent memo per reader that never takes another down (two readers built from the same spec each project once on the same landing and each keep handing out their own object); invalidating one base URL leaving every other base URL's reading untouched; `invalidateEngineCaps` being a no-op on an empty or non-string base URL; the retired per-reader `forget*Reading` exports being absent from the root entry; the read anchor being resolved dynamically; and the deliberate split in how presence is judged per reader (own data property only, so a key that exists only on the prototype is not reported). A stored reading is frozen at every level, whether it reports a value or reports that the position was not reported, so a consumer that writes to the object a reader hands out changes nothing for later readers (a strict-mode write throws); the capabilities body passed in is never frozen. A reading follows landings and invalidations rather than any per-reader state: after `invalidateEngineCaps` every reader answers "unobserved" until the next landing, checked against a late response from the previous engine (dropped by the cache), a replacement probe that succeeds, throws, returns a bad shape, hangs or never comes, two readers sharing one landing, two base URLs landing in either order, two in-flight probes landing in either order, and a projection that invalidates midway — and a test reset of the cache never revives an older reading. A host that never invalidates is checked on a seeded random sequence (each read is the projection over the latest landing, one object per landing, getters never executed), and the downstream decisions that depend on these readings are pinned at each moment around a switch. |
+| `scripts/run-engine-cap-reader-factory-test.mjs` | The one shared implementation behind every capability reader's ports, exercised as a table. Since 0.95.0 a reading is a lazily computed, frozen projection of the capabilities cache's landed response for that base URL, memoised per landing; a reader has a single read port — the probe tee port, the per-reader test reset and the per-reader `forget` port are all gone, invalidation is `invalidateEngineCaps` alone, and a test that wants a clean slate resets the cache. Every reader in the table runs the *same* criteria (the table length is the source of truth, and a roster check fails the gate if any source file calls the factory without having a row): the four states (nothing landed, present, key absent so not reported, malformed or unreadable once a response has landed, so a distinct unreadable state rather than unobserved); a throwing projection or descriptor read treated exactly like an unreadable one and never escaping; a projection that changes the generation mid-flight answering unobserved without being remembered; a getter at the top level or at any nested position never being executed and reading as malformed; a freshly minted "unobserved" reading on every miss (two misses are never the same object, so a consumer that mutates one cannot taint another base URL); one independent memo per reader that never takes another down (two readers built from the same spec each project once on the same landing and each keep handing out their own object); invalidating one base URL leaving every other base URL's reading untouched; `invalidateEngineCaps` being a no-op on an empty or non-string base URL; the retired per-reader `forget*Reading` exports being absent from the root entry; the read anchor being resolved dynamically; and the deliberate split in how presence is judged per reader (own data property only, so a key that exists only on the prototype is not reported). A stored reading is frozen at every level, whether it reports a value or reports that the position was not reported, so a consumer that writes to the object a reader hands out changes nothing for later readers (a strict-mode write throws); the capabilities body passed in is never frozen. A reading follows landings and invalidations rather than any per-reader state: after `invalidateEngineCaps` every reader answers "unobserved" until the next landing, checked against a late response from the previous engine (dropped by the cache), a replacement probe that succeeds, throws, returns a bad shape, hangs or never comes, two readers sharing one landing, two base URLs landing in either order, two in-flight probes landing in either order, and a projection that invalidates midway — and a test reset of the cache never revives an older reading. A host that never invalidates is checked on a seeded random sequence (each read is the projection over the latest landing, one object per landing, getters never executed), and the downstream decisions that depend on these readings are pinned at each moment around a switch. Both non-projection answers ("not observed yet" and "could not be read") are one frozen singleton shared by every reader, whatever the cause, so reading twice always yields the same frozen object; the answer shape itself is unchanged. |
 | `scripts/run-mcp-liveness-test.mjs` | The engine's **liveness observation** about each MCP server it hosts (`wiring_manifest.mcp[].liveness`, engine-side from core 7.24.3 / server 7.91.2): one reader, one word list, one leg-level verdict. The cell answers *can this server still be reached* — it is not the connect-time verdict beside it, which the engine deliberately freezes (a server that died mid-run still reads `connected`), and it is not a re-dial's judgement either, so `status: "failed"` next to `liveness.state: "reachable"` is a **real row**: the server answered the handshake and answered with a protocol error — up, and misconfigured. The three words are read as a closed set (an exchange completed / it was lost in transport or the clock / what came back does not answer the question), and a word outside it is malformed rather than rendered, because a word nobody upstream has defined is not a sentence worth putting on screen. **Absence is the fourth reading and is not one of the words**: it means *no liveness record is available*, which on the wire covers a server this leg never reached, a declaration that could not be dialled, an older engine, and a record the projection ahead of us dropped — all indistinguishable, so it is never read as "we looked and could not tell" (a strictly stronger claim), never as healthy and never as off. The failure-class footnote rides the unreachable word only, and one that turns up anywhere else, or that is malformed, loses **just the footnote** while the word and its timestamp stay: the honest reading is then "cannot be reached, reason not given", not "this record is broken". Malformed never becomes healthy: a cell that is present but unreadable marks its row and pulls the leg-level verdict back to *cannot tell*, since letting it sit beside a reachable row would report a leg as reachable on the strength of a record that may well have said the opposite. The verdict takes the worst fact first rather than a majority or the newest reading, carries no server count — so there is no fabricated zero to be read as "no problems" — and its timestamp belongs to **that leg's** observation, not to now: the engine runs no probe and adds no traffic of its own, so this is a per-leg snapshot rather than a heartbeat. The replayed roster on the session panel and the live leg go through the same reader, and the panel's own deployment-side rows carry no liveness position today, so the verdict does not borrow a word from that face. The word list is bitten in both directions where an installed witness exists and the absence of one is itself asserted against the installed engine's version, so the day it ships the comparison starts on its own; since sdk 11.2.0 the wire types declare the cell, and the word list and the presence test are pinned to those types at compile time in both directions, so the two can never drift apart |
 | `scripts/run-peer-lane-rules-write-capability-test.mjs` | Two more engine self-descriptions read the same four-state way as their seven sibling capability readers (`capabilities.peerLane`, `capabilities.permissionRulesWrite`): an absent key is not reported (an older engine that predates the position, never folded into `false`), `true` is present, `false` is a positive absent (the cross-session lane not being mounted on this deployment, or this particular call not being able to reach the tightening-direction write entry), and anything else is unreadable and drops the cell. Each carries its own single-source verdict (`peerLaneAvailable` returns `yes`/`no`/`unknown`; `permissionRulesWriteAvailable` collapses to a plain boolean, present being the only `true`). The write-entry position pairs with a boolean convenience port in the persisted-rules module, and this guard pins that port to derive from nothing but this one reader's own reading — never a conjunction with the lane-reachable position, and never a second read of the deployment-level existence signal the revoke surface uses (the two are documented as reading differently on purpose): a deployment where the lane answers true but the write entry's key is simply absent (an older binary) must still come back `false`, a deployment where the write entry answers true while the lane key is entirely unseen must still come back `true` (proving no silent conjunction crept in), seeding only the general capabilities cache — never this reader's own feed — must still come back `false` (proving the convenience port cannot be satisfied by the wrong table), and passing an explicit `undefined` base URL must still come back `false` even while a different, already-installed engine target answers `true` for the same position (an adversarial pass found the naive forward of that parameter falls through to the reader's own convenience default, silently answering for whichever engine happens to be installed rather than the caller's absent target — the fix routes an explicit absence through the same empty-string path the reader treats as unobserved). |
 | `scripts/run-lane-proof-identity-test.mjs` | The **instance identity of a lane proof**: the main-lane proof is minted fresh on every emission. Previously a single module-level constant object was handed both to `laneOf(an unregistered task id)` and to some fifty main-lane emission points, so two unrelated consumers — across adapter instances, across streams, across turns — held the same object: writing a card id onto one of them was readable on the other, and the four opening main-lane events changed together. Nothing in this package writes to a lane proof and the known consumers only read it, so this is an **aliasing hazard on a published output surface** rather than an observed corruption — a consumer that uses the proof as an identity key, for dedup, or as a view-layer identity would conflate two unrelated rows without writing a single byte, which is precisely the half that freezing the object would not solve. The gate therefore anchors on instance identity: two independent adapter instances, two rows inside one instance, the same id read twice, and two arms in one beat are each distinct references; mutating one leaves the others byte-identical; and the subagent lane, which already minted fresh, is the control that proves the criterion discriminates. The main-lane **value** is unchanged — an unregistered id still answers `{lane:"main"}` with exactly one own key and still emits its events, so absence is not turned into a second kind of absence — with ordering pinned three ways (registered-then-read, read-then-registered with no retroactive edit of an already delivered proof, the same id twice) and the id failure classes pinned four ways (unregistered, empty string, absent, non-string, the last two emitting no panel event at all rather than an ownerless proof). Where one row emits **two** events — the terminal-tick and card-close legs, which each yield a lifecycle stop and a panel end — the attribution is decided once (a consumer binding a card between the two yields must not split one row across two lanes) while each event still gets its own proof, so a host consuming them one at a time cannot poison the second before it is even yielded. The run stream leg is covered as the same shape, and a syntax-tree check forbids reintroducing a module-level lane-proof object literal or a module-level `LaneProof`-annotated binding (judged on the type node, not on text, so a compile-time pin tuple that merely mentions the type is not miscaught), backed by a type-checker pass that also catches an un-annotated module-level cache such as `const x = mainLane()` while letting the callable factory itself through, while the module-private three-state sentinels of the untrusted read are frozen instead — only `Object.freeze` counts, never `Object.seal`, which still permits writes to existing keys — their exposure being confined to one module |
@@ -427,14 +427,14 @@ guard still cross-checks the table by name).
 | `scripts/run-rewind-archive-capability-test.mjs` | The read face for whether a rewind can restore the code archive, and the honest three-state answer the ends render from it. The shells used to decide this from a local backup table that only their own in-process tools ever fill, so on any session where the engine runs the tools it stayed empty and the two code-restoring rewind modes simply never appeared, while the engine had been keeping a file history the whole time. The judgement now comes from what the engine itself advertises, and each of the four bits it advertises answers a different question: whether the conversation can be forked at a message at all, whether a fork can restore the tracked set, whether a code-only restore is possible, and whether this deployment understands the current spelling of the request key. The mode that rewinds the conversation and restores the code together needs both of the first two, and the engine offers no single bit for that combination, so the combination is made here: one bit stated off is enough to rule the mode out, both stated on make it available, anything else stays unknown — reading the file-history bit alone would offer that mode on a deployment that keeps file history but has no conversation anchors, where the request can only fail. A bit that is absent means an older engine that never spoke about it, which is not the same as an engine that said no, and a bit reported in a shape this reader cannot read is a third thing again — it is recorded as unreadable rather than quietly filed under "not reported", because those two send an operator to different places. One unreadable bit does not discard its siblings; only a response that is not a capability object at all clears the cell. What the ends get is available, unavailable or unknown, and unknown stays unknown: folding it into unavailable would hide the mode again, which is the mirror image of the bug this replaces. The sentences the doctor row can print are checked to be pairwise distinct and to avoid implying a refusal the engine never made, and the spelling of the outgoing request is deliberately not made to follow the epoch bit, since the older spelling is rejected outright by current engines |
 | `scripts/run-registry-quota-usage-test.mjs` | The projection of the cloud control plane's quota reading, and the three different things a missing number can mean there. This response says `null` in two places and means something different each time: no token quota is configured for this principal on this instance, and this window has no cap at all. Both are **facts the server is asserting**, not gaps in the reading — while a key that is absent or carries the wrong type is a genuine gap. All three have to survive to the screen separately, because folding them is how a user ends up staring at a confident `0`: an uncapped window rendered as if nothing were left, or a deployment that simply never configured quotas rendered as if the quota were exhausted. The complaint that started this was the opposite direction — a centrally configured quota that the command line could not see at all — so the reading also refuses to let an unreadable response masquerade as "no quota configured". Two fields deliberately do not share one signal: whether a window is exhausted and whether there is a recovery time, since the recovery time is only ever populated in the exhausted case and reading its absence as "not exhausted" would answer a question the response never answered. Counts that the server always provides are narrowed no further than the mint: a used counter has no uncapped state, so a null there is unreadable rather than zero. The wording helper carries the only human-facing phrasing, and the sentences for "no cap" and "unknown" are checked to contain no digits at all |
 | `scripts/run-file-history-capture-capability-test.mjs` | The engine's file-history-capture self-description (`capabilities.fileHistoryCapture`), read the same four-state way as its sibling capability readers: an absent key is reported as not reported (never folded into `off`), words are taken as an open set so a newer mode is not mistaken for a malformed answer, `fileHistoryCaptureMode` recognises `off` and `on-optional` (captured, and a run may opt out) — any other word, including the older `on-always` that pre-`7.104` engines sent, reads as unknown — and the wording for `off` speaks about capture only — whether code can be rewound is left to the rewind readings; a malformed reading makes neither the verdict nor the wording throw. The request side is covered too: the fragment helper yields the per-run opt-out word only when the reading is `on-optional` and the host asked not to capture — never for an older engine that would refuse the unknown key — and the fragment is driven through the real request constructor on every lane. The recognised words are checked both ways against the capture vocabulary of a `7.104` engine package. |
-| `scripts/run-model-identity-resolvability-test.mjs` | The model-identity judgement a client makes before letting anyone in: can the engine it is about to use start with a model name? Each end reports what it read from each place that can feed a model name to a local engine (complete, partial — a gateway address or a credential but no model name —, absent, or unreadable), or, for an engine that runs elsewhere, whether that engine has been seen answering; `modelIdentityResolvability` answers resolvable, unresolvable or unknown. Having part of an upstream configuration is not having enough of one, so partial lanes never add up to resolvable; a lane that was not reported or could not be read makes the answer unknown rather than unresolvable; an engine that runs elsewhere is never judged unresolvable and local lanes are never consulted for it (it does not start without a model name, so seeing it answer is enough to call it resolvable). `modelSetupDecision` combines that answer with whether this end can configure a model at all: setup is offered only for unresolvable on an end that can configure one, an end that cannot says so and points at whoever runs the engine, and unknown never opens setup. The detail and notice sentences are checked to be pairwise distinct, unknown sentences neither claim a model is configured nor that it is not, and the module is checked to import no platform I/O. The catalog lane counts as complete only when the host reports that the local engine accepts a catalog default; a report of `false` reads the catalog as partial, and no report (an explicit `undefined` included) or an unusable one leaves that lane undetermined — the same answer as before the catalog could be complete. Every sentence about a complete catalog names the catalog, including the undetermined sentence when the report is `false` and another lane cannot be read, and a 25-cell matrix pins that the report changes nothing when the catalog is not complete. |
+| `scripts/run-model-identity-resolvability-test.mjs` | The model-identity judgement a client makes before letting anyone in: can the engine it is about to use start with a model name? Each end reports what it read from each place that can feed a model name to a local engine (complete, partial — a gateway address or a credential but no model name —, absent, or unreadable), or, for an engine that runs elsewhere, whether that engine has been seen answering; `modelIdentityResolvability` answers resolvable, unresolvable or unknown. Having part of an upstream configuration is not having enough of one, so partial lanes never add up to resolvable; a lane that was not reported or could not be read makes the answer unknown rather than unresolvable; an engine that runs elsewhere is never judged unresolvable and local lanes are never consulted for it (it does not start without a model name, so seeing it answer is enough to call it resolvable). `modelSetupDecision` combines that answer with whether this end can configure a model at all: setup is offered only for unresolvable on an end that can configure one, an end that cannot says so and points at whoever runs the engine, and unknown never opens setup. The detail and notice sentences are checked to be pairwise distinct, unknown sentences neither claim a model is configured nor that it is not, and the module is checked to import no platform I/O. The catalog lane counts as complete only when the host reports that the local engine accepts a catalog default; a report of `false` reads the catalog as partial, and no report (an explicit `undefined` included) or an unusable one leaves that lane undetermined — the same answer as before the catalog could be complete. Every sentence about a complete catalog names the catalog, including the undetermined sentence when the report is `false` and another lane cannot be read, and a 25-cell matrix pins that the report changes nothing when the catalog is not complete. Since 0.100.0 the catalog lane takes the engine's adoption reading as a further input: `refused` reads a complete catalog as partial and wins over the older host report, while `undetermined`, `absent`, a missing key or an unknown word change nothing at all (verdict, sentence and wizard decision are byte-identical to not passing it); it acts only on a complete catalog lane, never on remote topology or an engine already answering, and its sentence names the catalog and differs from the older report's. |
 | `scripts/run-cloud-effective-projection-test.mjs` | The cloud control plane's effective-configuration response beyond its four configuration domains, and what a locally started engine does with the models document derived from it. Three top-level keys are read with the meaning their producer gives them: `warnings` (degradation warnings from the build that produced the served view — an empty list is a clean build, an absent key is an older server that cannot tell), and `budget` / `runtimeCaps`, where `null` means two different things: nothing resolves for this principal when you view yourself, and values withheld when the response previews another principal. A missing key or a wrong type is a third state, unknown, and none of the three is ever folded into a zero, a `false` or "no budget". A malformed warning row, budget field or cap costs only itself, and a known budget field of the wrong type is named as not shown rather than silently read as "no limit on this axis"; warning kinds are an open set, so a kind this client does not recognise still produces a warning line. The budget and cap readers are reconciled against the installed settings schema. The single wording source puts degradation warnings first and keeps every "not set / withheld / unknown" sentence free of digits, while a zero the server really sent is shown as a zero. On the models side, the document is written in the shape of the installed settings schema: the default model rides `roles.default` and the @-mention policy rides `atMention` (`only` carries its list, `none` is written, and `all` is left to the schema default). Every role goes through one gate: a role that names something outside the catalog is dropped and recorded, and a role that names a catalog model is carried only when the host injects an entry check that accepts that entry — one rejected entry costs only the roles that name it — while without the check no such role is written and each one is recorded with the reason `unvouched` (the missing check is the cause, not the role), because the package cannot tell which entries the local engine will discard and a role naming one makes it discard the whole models domain (a role chosen by capability or price names no entry and is always carried). When the host injects an entry check the document also carries the tier groups and the active tier group (without the check neither is written), and every catalog reference the local engine's schema would reject — a role, @-mention entry, tier binding or active group that names something outside the catalog served to this principal — is dropped and recorded, because one dangling reference makes the local engine discard the whole models domain and fall back to its environment catalog; this is proven by reading the produced document with the installed file store. The @-mention policy is narrowed with the settings contract's own narrowing function: an allowlist pruned to empty becomes "nothing may be mentioned" (never "everything") and gets one warning saying so, and a policy this client cannot read is not written (the schema default applies) and gets its own warning. The keys the settings schema retired (`default`, `atModelAllowlist`) are never written and, from 0.98.0, never read: a reply carrying them and the same reply without them give byte-identical documents, issue lists, notices and pool-delete answers. Both generations of the same response produce the same document, byte for byte, and the gate compares the result with what the settings contract's own one-time migration makes of the older shape. Warning kinds with their own sentence are reconciled against the kinds the installed schema can produce. A per-model budget in which no field could be read is never described as having no limits. Registry annotation keys on model entries (`origin`, `overridesTeam`) are removed before the models document is written: the local engine's schema does not accept them, and a configuration refresh would otherwise be rejected as a whole. A model entry the host-injected entry check rejects is left out of the document and references to it are dropped: the local engine drops such an entry at startup, but a refresh rejects the whole configuration over it, so the gate requires a clean read of the produced document; when no entry passes the check, the catalog is kept as sent and gets its own warning, which the gate proves by reading the document back. The check receives a copy, so it cannot alter what is written. Tier words outside the local schema's closed set are dropped and recorded as unsupported, and the package's tier word list is reconciled against the installed schema in both directions; an active tier group is judged against group names, never model names. Since 0.96.0 the MCP leg's account of HTTP transport keys is reconciled against the installed settings contract: the connector assertion header is not forwarded while the request shape has no such key (the guard turns red the day it does), and an entry that names both the principal header and the assertion header is left out with the reason `principal-and-assertion-header` instead of having one of the two picked, answering exactly as the contract's own parse does whenever both are well-formed names; the dropped entry's header references are not reported as missing environment names. The request shape is read with the type checker, so a quoted or inherited declaration counts as declared. From 0.98.0 an MCP http entry's `assertionHeader` is forwarded as written on both the cloud path and the local `.mcp.json` path, through one public reader that also leaves out an entry setting both identity headers, and a notice names every server that declares it, because no supported server sends that header yet. The written shadow models document is a data copy: changing the input afterwards, or a `model` getter that answers differently on a second read, does not change it. |
 | `scripts/run-websearch-verdict-test.mjs` | The per-request web-search configuration a host puts on the wire. Newer servers refuse the whole request when that section is malformed — and a missing or misspelled search provider now counts as malformed, because the section names where the searches go and an unknown destination is refused rather than silently swapped for the deployment's own backend. The old readers in this package dropped such a section without a word, which let the server swap destinations after all. The guard pins the new three-way verdict (absent, honoured, malformed with the field that is wrong) against the server's own judge, vector by vector, whenever that judge is available next to this package; it pins that a half-configured environment is malformed rather than ignored, that a malformed environment never falls back to the settings file (that would change the destination too), and that neither the sentence shown to the user nor the recorded reason repeats an endpoint, a key, or a search-parameter name or value — an unrecognised provider is never echoed either (the sentence lists the valid words instead), so a URL or key pasted into the wrong field does not come back out, even when it happens to be all letters. A host's key store is plugged in through a callback the package calls only after the provider has been recognised, so the precedence between environment and settings stays inside the package. Since 0.83.0 an endpoint that carries a user name or password is malformed as well (the server judges the same way from 7.101.0), the reason sentences match the server's own word for word, and the three older readers that dropped a misspelled provider are gone. Since 0.93.0 the search key is checked in the same pass: a key that cannot be carried as an HTTP header value — a line break, a NUL or another control character, or a character above U+00FF anywhere but at the very ends — is malformed, with the server's own sentence (the server refuses such a key from 7.118.0 on, and a deployment's own start-up configuration is refused too), while a trailing or leading newline, as secret files often carry, is accepted just as the runtime accepts it. The rule is a pure re-implementation of the runtime's two-step header check, so the package stays free of Node built-ins; the guard pins it character by character against the real Node header pipeline, sends the whole Latin-1 range through a real loopback request, compares a large corpus with the previous release byte for byte (legal keys read exactly the same; only keys that could never be sent differ), and pins that the server's refusal is read by its code alone — nothing in the package matches its sentence. |
 | `scripts/run-hooks-merged-disable-projection-test.mjs` | The fourth governance leg of the hooks projection: `disableAllHooks` set in a non-managed settings source. The value that counts is the **merged** one, read from the host through the optional `SettingsPort.mergedDisableAllHooks()`, because a per-source approximation ("any source says true") reads user `true` with local `false` backwards — the merged value there is `false` and every source's hooks ship. When the merged value is `true`, only managed-settings hooks are sent to the engine: non-managed settings can switch off their own hooks, never the managed ones, and a managed `disableAllHooks` is still judged first and sends nothing at all. A full matrix over the four sources, each true, false or absent, is merged with the reference rule (later sources override earlier ones, managed settings last) and every cell's projection is asserted. The reader is the only authority: per-source values never second-guess it, and only a strict `true` counts. A host that does not implement it keeps the previous behaviour and gets exactly one warning per installed settings port, never one per request, and none on paths where the reader would not have been consulted; a reader that throws is treated as `true`, so managed hooks still ship. The session goal's Stop hook and the final-verification yield rule, which reads the projected hooks, follow the same verdict, and the trust gate and the three managed gates are evaluated before the reader is ever called. The last leg pins the member's declared shape in the built declarations: optional, no parameters, returning a boolean or `undefined`. The flag settings source (a settings file or inline settings given at startup) is projected after the local source, and settings hooks stay concatenated managed first (managed → user → project → local → flag): the engine runs hooks in request order, and three of its budgets go to whoever comes first — the per-event wall-clock budget, the per-event cap on model-backed entries, and the total cap on added context — so a managed hook placed after the others could be crowded out and silently not run. The flag source moves with user, project and local under every managed or merged gate; its exec-form entries are dropped and warned about like any other settings source, and the not-run notice names it `Flag settings`. An optional leg feeds the request body to an installed server's hook runner and requires the managed deny, block and context to take effect while non-managed hooks exhaust each of those budgets, and pins the known cost of that order: a non-managed `PreToolUse` hook placed after the managed one can still rewrite the input after the managed check allowed it. Without an installed server the leg reports that it did not run. Safe and bare mode: the host reports its startup mode through the optional `SettingsPort.hooksStartupMode()`, which both the plain and the plan path read and which answers `'safe'` or `'bare'`; the flag already carried by the plugin hooks reading, which only the plan path reads and which cannot tell the two modes apart, counts as safe mode unless the reader named a mode. In either mode only managed-settings hooks are sent from the settings sources: bare mode is handled like safe mode, so managed hooks are still sent, also under the managed-only gates. In both modes plugin hooks are excluded as before and the session goal's Stop hook is still sent. Only the two exact words count; a host that does not implement the reader, returns `undefined`, returns any other value or throws keeps the previous request byte for byte, checked for every such reply, both flag states, both paths and with or without a session goal; any other value and a throw each leave one debug line, and so does a reader written as a property instead of a method, which narrows nothing. The reader is called at most once per request and not at all when hooks are already switched off entirely. The optional server leg requires the left-out sources' hooks not to run while the managed ones still run and still deny, in either mode. A hook written in more than one settings place is sent once: entries on the same event, in groups that are identical apart from their hooks (same matcher), with the same identity — command, shell, arguments and condition for command hooks; type, prompt and condition for prompt and agent hooks; URL for HTTP hooks; server, tool and input for MCP tool hooks — are sent once, at the position of their first occurrence in the concatenation order and with the fields of the copy from the highest-precedence source (managed, then flag, local, project, user; within one source the later copy), so a managed copy keeps both its place and its values and a user copy duplicated in flag settings keeps its place but uses the flag copy's timeout; the other entries keep their order, groups without duplicates are sent as they were, and a group left empty is not sent. Different matchers, events, a one-byte command difference, a different condition, an absent versus explicit shell, different types or extra group keys are not merged; malformed entries and entries whose identity cannot be computed are left alone; plugin groups and the session goal hook do not take part, and hooks removed by safe or bare mode never come back. The optional server leg requires such a duplicate to run once with its added context appearing once, and a user copy with a one-second timeout duplicated by a flag copy with a ten-second timeout to run to completion on a two-second command. |
 | `scripts/run-memory-saved-projection-test.mjs` | Engine memory writes (a successful `Remember` tool call) moved off the transcript onto the additive `memory_saved` chrome event, driven through the real pipeline: zero transcript rows for the write (the transcript is byte-identical to the same frames with the write reported as not successful), exactly one event whose `notes` carry the note text verbatim (notes, not file paths) and whose key set is exactly kind / laneProof / id / notes; no event for a missing, empty or non-string note, a non-`true` `ok`, a tool error, a missing result or another tool name; two writes give two events in order with distinct ids; a sub-agent write rides the sub-agent lane and an empty parent id emits nothing rather than falling back to the main lane; the `id` is derived from the write's wire key (the tool-end event id, else the tool-start event id, else the call id; empty ids count as absent), so projecting the same wire events twice gives the same id, and it never collides with the tool result row of the same or another call; the event sits right after the tool result row; the arm is registered as required. |
 | `scripts/run-result-frame-projection-test.mjs` | Result frames and the synthesized terminal rows. The CC key `terminal_reason` is minted on result frames only where it follows from what the engine reported: `completed` on success, `max_turns`, `budget_exhausted` and `structured_output_retry_exhausted` for the three matching engine codes, on both the done-frame path and the failed-event path. Every other outcome leaves the key absent as an own property rather than present with an undefined value: wall-clock and token-budget limits, the classifier denial limit, cancellation, unknown codes, blocked, paused, unreadable or missing terminal records, and the busy-session refusal. The public reader `terminalReasonForResult` shares the minting predicate and is checked to agree with the minted key on every frame the gate produces. Both the minted key and the reader derive the word from the frame's CC subtype (success with `is_error` strictly false, and the three limit subtypes), not from the error code, so a replayed row whose status is paused, blocked or unrecognised never carries a word that contradicts its subtype. The four words are checked against the mirrored CC union, and the minting file is checked to hold no hand-copied code literals. The renamed superset keys (`_sema_error_code`, `_sema_salvaged_result`, `_sema_model_degraded`, `_sema_selected_model`, and the row flag `_sema_api_error_message`) are driven through the real stream pipeline. Each must be present under its new name, the old name must be absent, and every frame the gate saw is swept for old names. The selected model appears on error envelopes whenever the terminal record carries it, and never on a failed event, which has no record. It stays separate from the provider-reported model name. The two in-package readers still work: the interactive result arm reads the salvaged text under its new name, and the print init gate treats the renamed flag as the run having ended. Since 0.86.0 the result arm no longer reads `result` on an error envelope, meaning a frame whose subtype is a string starting with `error_`, including error subtypes it does not recognise yet; a compile-time check pins that every CC error subtype carries that prefix. Such an envelope that carries the salvaged text only under the old `result` key yields no answer row, no appended suffix and no divergence event; when both names are present, or the new one is present but not a string, the old one is never read. Everything else reads `result` exactly as before, in all three producing cases (whole answer, missing suffix, divergence): success frames, a success frame flagged `is_error: true` (checked equal to the same frame with `is_error: false`), and frames whose subtype is a string without that prefix (empty, unrecognised, `Success`, a bare `error`), absent, or not a string (each checked equal to the same frame with subtype `success`). From 0.94.0 four root readers return the result frame's string superset keys (`resultErrorCodeOf`, `resultSelectedModelOf`, `resultSalvagedResultOf`, `resultParkFailedReasonOf`): they read only the frame's own data properties, never fall back to the retired names, run no getter and never throw. A synthesized terminal row that carries `_sema_api_error_message: true` has its text credential-washed on both projection paths (minted by the stream projection, or rebuilt by `adapt()` from an older transcript row), so the flag itself tells a client the text is already washed; model rows are left byte-for-byte alone. |
 | `scripts/run-layering-shadow-export-test.mjs` | Same-name shadows across the first-party clients that consume this package (terminal, desktop, web and the admin console). Each client's product sources are read at the local clone's `origin/main` (or its HEAD when there is no such ref), without fetching, and parsed with the TypeScript parser; every top-level runtime export the client declares itself is compared with this package's public runtime exports. The guard prints which ref, commit and commit date it read for each client, and warns (without failing) when that commit is more than seven days old, because the result then only describes that older snapshot. A client-side declaration carrying the name of a package export means a piece of shared logic now lives in two places and can drift apart. It fails the guard unless it is listed in `scripts/layering-shadow-exemptions.json`, and a listed row must carry one due-by version no more than three minor lines ahead (it fails once the package reaches it; that single field is the only one the guard reads for expiry, and the older retire-by and review-by fields that still sit on the rows for one more version are only checked to agree with it). It also fails once the client has removed the shadow and the row still stands. Every row also names what kind of duplicate it is, and carries the evidence that kind requires. A client copy that is this package's own function object behind a type assertion, or a thin wrapper that only passes the client's own dependencies into this package's function, must be proven so on every run by reading the client's syntax tree; such rows are due for review within two minor versions, and the guard fails as soon as the proof no longer holds. A wrapper whose extra logic the client has confirmed to be host-specific carries that confirmation (the client's own wording and where it was stated) together with a pinned review record that is due within two minor versions. A wrapper that adds its own decisions, or a same-name function that does something else, carries a review record pinned to a hash of the normalized client declaration (comments and formatting do not count); once the client's copy changes, the guard fails until it is reviewed again. A second implementation of the same logic carries a semantic diff: the client's copy is taken from the commit being read, the named declarations and the client modules they import are extracted with the TypeScript parser, transpiled and run in memory against this package's build on the same inputs, and any disagreement outside the classes registered on that row fails the guard (each class is a fixed predicate over both answers). A separate register lists same-meaning copies that carry a different name (a name-based comparison cannot see those): each row names the client declaration and the package export it duplicates, and fails when it is malformed, past its due-by version, or when the client no longer declares that name; the register only records and checks liveness and expiry, it does not scan for such copies. A row can follow a client-side rename to a new name, and can be registered ahead of a client change for a limited number of versions before the client code exists. Re-exports of this package's own exports are the intended form and never count. A client tree that is not present is reported as a skipped section, not as a pass. The ruler proves itself on an in-memory fake client (planted shadows must be caught, legal forms must not), on a throwaway repository (a missing `origin/main` falls back to HEAD, a broken one is a fault rather than a silent fallback), and refuses to report zero on a client whose scan surface is empty. |
-| `scripts/run-session-policy-deliverable-test.mjs` | Which of a batch of user-written permission rules can be written into a session’s own rule record without changing their meaning, and why each of the others cannot. The record holds whole tool names and command names only, so exactly one class maps across losslessly: a deny rule that names one tool with no qualifier. Everything else is withheld with one word from a closed seven-word list — an ask rule (the record has no ask tier), a deny rule with a parenthesised qualifier (recording just the name could block more), a rule that names a server or agent peer without naming one of its tools (for every protocol namespace the engine knows, checked against the engine package's own table) or contains a wildcard (*) anywhere (an engine that compares exact names would block nothing), an entry that is not a tool name, and a name the engine refuses at the start of every run — a retired tool name, or one containing "__" without a protocol prefix, where the prefix check is case-sensitive (once such a name is in the record, every later run of the session fails at startup until that entry is removed; the retired-name list is checked entry by entry against the engine package's own list, and a withheld retired name carries its current name when the engine says it was renamed), and — only when the caller passes the tool roster of a run — a name that roster does not list as a tool name or alias, compared exactly with letter case counted (recorded as it is it would block nothing on that deployment) — and each word has one sentence, which never echoes the rule itself; asking for the sentence never throws, even with a value that throws when turned into a string. A name with leading or trailing whitespace counts as not a tool name: the record compares exact bytes, so it would block nothing. The guard pins the batch semantics: the deliverable part is either the whole batch or empty, never a subset, so a caller cannot send half a change and report it as saved. An end-to-end check runs the engine package itself: every batch this function would deliver — the recorded vectors and a fixed-seed sample of generated names — is written into an in-memory session rule store and the next run must get past its start-up checks and reach the model, while every name withheld as refused — every retired name included — must indeed make that run fail at start-up, and every string literal in the judgement source that it withholds as refused must be one the engine package's own tables refuse. It also checks that malformed input never throws and never delivers anything (non-arrays, non-string entries, holes, a polluted array prototype, a length or index that throws, a changing index read once), that a batch which cannot be read at all is marked `unreadable: true` while an empty batch is not, so the two stay tellable apart, that each word is produced by some vector and nothing outside the list is produced, and — when a checkout of the previous in-client implementation is present — that this function gives the same answer on every recorded vector and on tens of thousands of generated rules and pairs, except for five deliberately stricter classes (whitespace-padded names; rules with a wildcard anywhere, which the previous implementation sent as exact names unless the wildcard was the whole tool part of a server rule; peer-wide rules outside the MCP namespace, which it did not recognise; names the engine refuses at start-up, which it sent as ordinary names; and names the engine's rule compiler does not accept as one whole tool, such as an escaped parenthesis or a backslash in the name), whose disagreements are counted per class and must match an independent count exactly. The optional roster only ever withholds more: with no roster, or an absent or null one, every answer is byte-identical to the last release before the roster existed, checked against that release's published file over tens of thousands of inputs, apart from the rule-syntax class described below; a name withheld without a roster stays withheld with the same word, and the same current name, when the roster lists it as a name or an alias; a name delivered without a roster is withheld as not in the roster exactly when the roster lacks it; the engine's own namespace-covering forms keep their word; and a roster or options value that cannot be read marks the whole batch unreadable instead of falling back to no roster, while a readable empty roster is a roster. An end-to-end check takes the roster from a real run of the engine package that mounts a tool with an alias: every name delivered with that roster passes the engine's start-up name audit without being reported, and every name withheld as not in the roster is one that audit reports as matching no mounted tool. When the previous in-client implementation carries its own roster check, the two are compared under a given roster and may differ only for an alias (delivered here, withheld there) and for a refused name that the roster lists (withheld here, delivered there). A deny rule is delivered only when the installed engine's rule compiler, asked about that one string the same way the request-body permission assembly asks about whole tool names, accepts it as a rule for one whole tool: a name with an escaped parenthesis, a backslash, a lone closing parenthesis or more characters than a rule may have is withheld as not a tool name, because the record would compare those bytes and match no tool. When the compiler is not installed or cannot answer, a batch that needs it is marked unreadable rather than delivered, while batches that never reach that step answer as before. The guard checks that every delivered name is one the compiler accepts, that nothing delivered now was withheld by the earlier release and the extra withholding is exactly that class, that a roster never makes such a name deliverable, that the compiler of every engine generation in the server fixtures gives the same answers, and that the request-body assembly reaches the same verdict on the same names. |
+| `scripts/run-session-policy-deliverable-test.mjs` | Which of a batch of user-written permission rules can be written into a session’s own rule record without changing their meaning, and why each of the others cannot. The record holds whole tool names and command names only, so exactly one class maps across losslessly: a deny rule that names one tool with no qualifier. Everything else is withheld with one word from a closed seven-word list — an ask rule (the record has no ask tier), a deny rule with a parenthesised qualifier (recording just the name could block more), a rule that names a server or agent peer without naming one of its tools (for every protocol namespace the engine knows, checked against the engine package's own table) or contains a wildcard (*) anywhere (an engine that compares exact names would block nothing), an entry that is not a tool name, and a name the engine refuses at the start of every run — a retired tool name, or one containing "__" without a protocol prefix, where the prefix check is case-sensitive (once such a name is in the record, every later run of the session fails at startup until that entry is removed; the retired-name list is checked entry by entry against the engine package's own list, and a withheld retired name carries its current name when the engine says it was renamed), and — only when the caller passes the tool roster of a run — a name that roster does not list as a tool name or alias, compared exactly with letter case counted (recorded as it is it would block nothing on that deployment) — and each word has one sentence, which never echoes the rule itself; asking for the sentence never throws, even with a value that throws when turned into a string. A name with leading or trailing whitespace counts as not a tool name: the record compares exact bytes, so it would block nothing. The guard pins the batch semantics: the deliverable part is either the whole batch or empty, never a subset, so a caller cannot send half a change and report it as saved. An end-to-end check runs the engine package itself: every batch this function would deliver — the recorded vectors and a fixed-seed sample of generated names — is written into an in-memory session rule store and the next run must get past its start-up checks and reach the model, while every name withheld as refused — every retired name included — must indeed make that run fail at start-up, and every string literal in the judgement source that it withholds as refused must be one the engine package's own tables refuse. It also checks that malformed input never throws and never delivers anything (non-arrays, non-string entries, holes, a polluted array prototype, a length or index that throws, a changing index read once), that a batch which cannot be read at all is marked `unreadable: true` while an empty batch is not, so the two stay tellable apart, that each word is produced by some vector and nothing outside the list is produced, and — when a checkout of the previous in-client implementation is present — that this function gives the same answer on every recorded vector and on tens of thousands of generated rules and pairs, except for five deliberately stricter classes (whitespace-padded names; rules with a wildcard anywhere, which the previous implementation sent as exact names unless the wildcard was the whole tool part of a server rule; peer-wide rules outside the MCP namespace, which it did not recognise; names the engine refuses at start-up, which it sent as ordinary names; and names the engine's rule compiler does not accept as one whole tool, such as an escaped parenthesis or a backslash in the name), whose disagreements are counted per class and must match an independent count exactly. The optional roster only ever withholds more: with no roster, or an absent or null one, every answer is byte-identical to the last release before the roster existed, checked against that release's published file over tens of thousands of inputs, apart from the rule-syntax class described below; a name withheld without a roster stays withheld with the same word, and the same current name, when the roster lists it as a name or an alias; a name delivered without a roster is withheld as not in the roster exactly when the roster lacks it; the engine's own namespace-covering forms keep their word; and a roster or options value that cannot be read marks the whole batch unreadable instead of falling back to no roster, while a readable empty roster is a roster. An end-to-end check takes the roster from a real run of the engine package that mounts a tool with an alias: every name delivered with that roster passes the engine's start-up name audit without being reported, and every name withheld as not in the roster is one that audit reports as matching no mounted tool. When the previous in-client implementation carries its own roster check, the two are compared under a given roster and may differ only for an alias (delivered here, withheld there) and for a refused name that the roster lists (withheld here, delivered there). A deny rule is delivered only when the installed engine's rule compiler, asked about that one string the same way the request-body permission assembly asks about whole tool names, accepts it as a rule for one whole tool: a name with an escaped parenthesis, a backslash, a lone closing parenthesis or more characters than a rule may have is withheld as not a tool name, because the record would compare those bytes and match no tool. When the compiler is not installed or cannot answer, a batch that needs it is marked unreadable rather than delivered, while batches that never reach that step answer as before. The guard checks that every delivered name is one the compiler accepts, that nothing delivered now was withheld by the earlier release and the extra withholding is exactly that class, that a roster never makes such a name deliverable, that the compiler of every engine generation in the server fixtures gives the same answers, and that the request-body assembly reaches the same verdict on the same names. Since 0.99.0 the cells that depend on the engine's tool-name grammar take their expectation from the installed engine generation, which refuses blank, invisible and wildcard-only tool names. |
 | `scripts/run-plugin-hooks-projection-test.mjs` | Plugin hooks: each command hook an enabled plugin declares is decided one by one as running in the engine, running in this client, or not running at all, and the page of hooks sent with a request is built from the same per-turn plan the client uses to skip its own copies, so one hook never runs in two places. Governance is judged first and always wins — a managed hooks switch-off, an untrusted workspace, safe or bare mode, or a governance read that fails sends no plugin hook and does not list it as a gap; managed-hooks-only (set directly, or through a merged non-managed hooks switch-off) keeps only managed plugins; the plugin-only customization lock does not touch plugin hooks. A hook reaches the engine only when this client started the engine on this machine, the engine reports plugin-hook support, the entry is a command, the plugin declares no sensitive option, and the event still fits the engine's per-event limits; the gate walks that matrix cell by cell, including the limit boundaries and a session goal hook counting toward them. A fact that was never read is reported as not known rather than as a fact: a host that does not say where the engine runs gets a "not known whether this client started the engine" reason, an engine whose capabilities have not been read yet gets a "not known yet whether it supports plugin hooks" reason, and the plan's two engine facts are null in those cases, not false. Events the engine never fires run only if the client says it fires them itself, and hooks the upstream behaviour itself refuses (option references in a shell-form command, an unset option in exec form, malformed entries) run nowhere. Exec-form arguments are passed element by element with only saved non-sensitive option references filled in; path placeholders are left for the executor. Sensitive option values never reach the request: with a host that wrongly supplies one, every string in the plan, the request body, the notice, the labels and the log is searched for it across eight cells. A host without the plugin reader keeps the previous request body and gets exactly one warning per settings port; plugin data that throws while it is being read (a throwing getter, a revoked proxy) is treated like a failing reader — no plugin hooks this turn, settings hooks still sent, nothing thrown; the not-running notice names the plugin and events, never a command or an option value, and escapes control characters in names. Command hooks from settings that carry arguments (a non-empty `args` array, which is the exec form, or any other non-null value) are removed from the request until the engine reports support for arguments, because the engine would otherwise drop the arguments and run the bare command through a shell; an empty `args` array is not treated as carrying arguments when the command is made only of letters, digits and `_ . / : + -` (the shell runs the same executable), so such a guard still reaches the engine, while an empty array on a command with spaces or shell characters is removed; `args` on a prompt or http entry, a null `args`, or an entry with no type is left alone, and those go out unchanged. MCP tool hooks, which the engine cannot parse, are removed from every request: a plan records each one and its not-running notice lists them — the gate checks request bodies against the engine's own hooks schema, and the raw settings against it as a positive control. Malformed entries still pass through for the engine to reject loudly. Since 0.98.0 the page of hooks is built only from a plan: `hooksForWire` takes `{ plan }` and throws a TypeError for anything else (nothing passed, null, a value that is not a plan, a near-plan, a plan passed directly, an options object whose `plan` cannot be read, or a host value that throws while it is checked), naming the right call and never echoing the value or logging what the host value threw; the gate walks those shapes one by one and checks the built output for leftovers of the retired path. A `plugin` key written by hand on a settings hook is stripped before sending (even when its value is undefined), because only hooks that come from the plugin reader may carry plugin context; the settings document itself is left untouched and a debug line records the count. The two hand-copied tables, the engine-fired event list and the engine limits, are checked against their owners. Since 0.89.0 the same per-turn decision also gives every settings hook on an engine-fired event exactly one executor: a command, HTTP, prompt or agent hook from settings goes to the engine when the engine reports that it runs hooks sent with requests and either this client started the engine on this machine or this client cannot run that hook itself; otherwise this client runs it and the hook is left out of the request. A hook the engine is known to skip (a PowerShell command, or hooks the engine ignores entirely) that this client cannot run either is listed as not running; one the engine may or may not reach (capabilities not read yet, or a prompt or agent hook after the fourth one on an event) is still sent and not listed. The client learns which of its own settings hooks to skip from `settingsHooksKeptLocally`, which matches hooks by content against the request body only, returns the same array when nothing is skipped, and for the two compaction events follows whether the engine actually compacted. When the prompt was not written by the user, prompt-submit hooks from settings and plugins are left out on both sides. A host that reports events must also report which hook types its own executor runs on each event: with that table, a plugin hook it cannot run is reported as not running rather than as running in this client; without it, prompt and agent hooks from settings on engine events are treated as hooks the client cannot run and stay in the request for the engine. A settings hook whose required field is unusable (a command, URL or prompt that is not a non-empty string) is not assigned to either side and is sent unchanged, so the service still rejects it; malformed rows in a plan handed back by the host never make the not-running notice throw and never authorise skipping a hook; and a missing, empty or `*` matcher counts as the same match-all matcher when the client decides what to skip. A matrix over source, event, hook type, capability, engine ownership, host shape, prompt origin and compaction outcome checks that every hook runs at most once and exactly once whenever either side can run it, while the print lane and hosts without a local executor keep the previous request byte for byte. An optional leg runs the request body through an installed server's hook runner, current and previous release, together with the client-side skip, and counts one run per prompt. A plan in the previous release's six-key shape is read the same way by the request side and the skip side, so every hook it sends runs once; the client side of the skip reads groups and entries the way the client's executor does (class instances and inherited getters included); and on the two compaction events a prompt or agent hook past the engine's per-event limit runs in the client when the client can run it. |
 | `scripts/run-display-untrusted-projection-test.mjs` | The single display-safety outlet (`displayUntrusted`) and the credential wash on the end-of-run rows this package mints. The outlet composes two credential nets (URL structure: userinfo, every query value, the fragment, path parameters and path segments that start with a known secret prefix; key/value words such as `Authorization: Bearer ...`, `Authorization: token ...` or `api_key=...`, plus well-known secret literals that appear without a label, such as `sk-...`, `ghp_...`, `AKIA...`, JWTs and the body of a PEM private key) with three character nets (control characters, bidirectional and format characters, whitespace folding). The credential nets match on a view of the text with ANSI sequences, format characters, control characters and the outlet's own escape tokens stripped, and map the result back onto the original, so colouring or an invisible character wedged between a label, its separator and its value cannot hide the value, and no stray marker is left behind. Whitespace of any length around the separator is accepted. Hosts, ports, paths, query key names and surrounding prose stay byte-for-byte, clean text comes back unchanged, the result is idempotent (also with a length cap), a length cap never splits an escape token or a surrogate pair, an invalid cap means no cap, and every net can be switched off on its own. A few narrow shapes are left alone because they name something rather than carry a value (a plain English word after `bearer` or `basic`, a back-quoted credential variable name, a plain integer after `tokens:`, a list of key names after `keys:`), each with a counter-example that is still washed. Regional flag emoji built from tag characters are kept whole. The existing single-line helpers (`escapeDisplayControlChars`, `collapseLabel`, `capForDisplay`, peer sender names and the hook failure banner) run on the same engine and, since 0.96.0, read the same character table as the outlet: each is checked equal to the outlet with both credential nets off (escape without folding, fold then escape, space mark then fold, and the same length cap) over a seeded pool of more than 600 strings; every character class the old sets left alone (zero-width characters, the word joiner, invisible operators, U+FEFF, the soft hyphen, tag characters and other format characters, plus C1 controls, bidirectional controls and lone surrogates in sender names) is marked class by class, while joiners inside emoji sequences and joining or Brahmic scripts and RGI regional flags stay as written; no character the old sets marked is ever left raw, input without the newly marked characters stays byte-identical to the previous output over every BMP code unit plus random strings, all of them are idempotent, and neither `capForDisplay` nor the hook banner's 512-character cap splits a `\u{…}` escape token or a surrogate pair. The approval decision-note echo, the subagent resume receipt (and its failure debug line) and the startup list of plugin hooks that will not run now also drop bidirectional and format characters (and, for the receipt, C1 controls); a note that is empty after cleaning is treated as absent. The synthetic end-of-run rows (`API Error:`, `Run stopped:`, `Model output error:`, `Outcome unknown:`) and the result frame's `errors[]` pass both credential nets before they leave the package, on the print lane and on the interactive lane (which also keeps the row-class flag); this covers a blocked reason whoever wrote it, while assistant text rows, a successful `result` and salvaged output are never touched, and a non-string `errors[]` entry is passed through unchanged. The known-secret-prefix check is a local copy of the configuration package's detector and is compared with the installed one entry by entry. Since 0.83.5 the outlet has two opt-in switches and a position read-out. `escapeBackslashes` (escape form only) writes every literal backslash as a pair, so each output decodes back to exactly one input (a real invisible character and its literal six-character spelling no longer look alike); a reference decoder round-trips thousands of random strings, the output is byte-identical to the default when the input has no backslash, a length cap is measured on the paired output and keeps the longest fitting prefix, credentials are washed exactly as in the default, and the switch is not idempotent by design (use it only at the final render). Zero-width joiners and non-joiners are kept only inside emoji sequences drawn as emoji (so not between symbols such as © or ™ that display as text) and between letters of scripts where they change the shaping (joining scripts such as Arabic, and the Brahmic family), each listed script checked both ways; the one other place a joiner is kept is right after a virama at the end of a word, the older spelling still found in Malayalam and Bengali text. Next to Latin, Cyrillic, CJK and other letters, next to modifier letters shared across scripts, at the start of a word, at the end of a word without a virama before it, or on their own they are now marked. `blanks` marks characters that look like a space but are not an ASCII space (no-break and other width spaces, the ideographic space, Hangul fillers, the blank Braille pattern) before whitespace folding, for names that must never look alike. `displayUntrustedMarks` returns the same text plus the position of every character mark; its text is compared with the outlet over thousands of inputs. With both credential nets off, the character face stays byte-identical to the previous release for input without joiners. The credential nets read escape sequences the way a terminal would when one is cut short: an unfinished colouring or character-set sequence interrupted by another one is dropped as a whole, a sequence never takes the `@` of an address as its final character, and a final character that starts a well-known secret literal (`sk-`, `ghp_`, `AKIA`, a JWT) is also read as the start of that literal; escape tokens this outlet writes are read as one unit, while look-alike text it never writes (an upper-case `\U`, or a code point it never marks) is read as plain text. A URL is cut before a run of non-ASCII blanks followed by a credential label or scheme word, Hangul fillers and the blank Braille pattern count as spaces around a label's separator, and a value that itself starts with a quoted label (`token= "password":"..."`) is left to that inner label. With `blanks` on, a blank written as an escape token right after a label is read as a blank when the value is judged, so `password:` followed by a no-break space and `missing` stays unmasked and an empty value gets no marker; the credential nets also read the text the way it looks after default whitespace folding and combine what each reading masks, so whatever the default form masks stays masked with `blanks` on (checked over a seeded corpus for the escape form, with paired backslashes and without folding; the exceptions are text that itself contains a literal six-character blank escape, which cannot be told apart from one the outlet wrote, and the dot and space marks, which cannot tell a marked blank from a real dot or space). A lone surrogate wedged between a label, its separator and its value no longer hides the value: the credential nets treat it exactly like a format character, in the machine-readable wash, in a single pass of the display outlet, and on the end-of-run rows and the result frame's `errors[]` on both lanes, while lone surrogates anywhere else are left byte-for-byte. Since 0.84.1 an address-shaped value (`scheme://…`) that sits right after an invisible single character — a format character, a lone surrogate, a non-whitespace control character, or the outlet's own escape token for one — is no longer let through as an address: the scheme stays and the host and path are replaced, with the query and fragment masked as before; `user:<password>@` followed only by stripped units and then a boundary, a port or another `@` is masked as userinfo. A value after real whitespace or after a colour sequence is still treated as an address, and clean text stays unchanged; forms that the previous release masked are checked not to leak on the same outlets. A `user:<password>@` candidate that is the value of a credential label or scheme word — including a label or scheme word split by stripped characters, and a quoted value that goes on past whitespace — is left to the label pass, so the whole value is masked as in the previous release; both reported shapes and frozen samples from the targeted pools are checked on every outlet. A scheme-less `user:password@host` whose password holds a double quote, backtick or angle bracket is still masked when a credential label or scheme word sits right before the user name, or when the user name is a plain word that is not itself a label, the host has a real shape and the address is not wrapped by that quote; quoted JSON values, quoted or bracketed e-mail addresses and quoted arrays stay byte-identical. When a label inside a quoted value ends right at that value's closing quote, the quote and separator stay outside the mask, and an address that begins the rest of a quoted value counts as an address only up to the next blank, so washing the output again never changes it. The credential stage now repeats its single scan on its own output until nothing changes (the cap counts rewrites, at most eight; clean text takes exactly one pass), so masking the output again never changes it for every input that settles within the cap — the random, exhaustive and grammar pools and the two known step-by-step constructions settle in at most three passes counting the confirming one, and the searched constructions kept in the pool in at most six; a deliberately constructed input can need more rewrites as it grows, and at the cap the last rewrite is returned, which a further pass can only mask more of. The checks cover the pass cap and how reaching it is reported, single-pass clean text, a grammar sample of the busiest shapes, constructions that once advanced one step per pass, the mark positions on the final output and how positions compose across rewrites. A username that is itself a credential label with a quote or angle bracket in the password is masked as userinfo (also with only the URL net on) unless the label's own value really reaches the `@`; a URL parameter such as `;token=...` or `,token=...` stays inside the URL piece; a private-key header the outlet cannot recognise does not hide the labels inside its block; the tail glued after the host of a label's address value is masked to the value's end; and a value this package's own transcript wash split at a carriage return is masked when the display joins it back. A bare `Token <value>` / `token <value>` (no label, no separator) is now washed like `Bearer`, with a wider allow-list of ordinary English words after `token` so prose such as `token count: 1234` or `token expired` stays readable; a `/token` URL path tail followed by a status code is not a scheme word and stays as is, and neither is `token` right after `unexpected` in a parser error such as `SyntaxError: Unexpected token '<'`, which stays readable. The `token` scheme word only adds masking of its own value: label values and `bearer` / `basic` values are read exactly as in the previous release, so a value that starts with another character, or a `bearer` value, followed by `token`, a non-ASCII space and an address is still masked whole. When a credential label or scheme word is followed by a colon and a YAML block scalar header (`\|` or `>`, optionally with `-`, `+` or digits, invisible characters allowed between them, and nothing else on that line but blanks, invisible characters or a comment, ending in a line break), the whole indented block becomes a single marker: the block runs until the first line with visible content that is not indented deeper than the label's line, blank lines included, and the line break before that line and the line itself stay as they were on outlets that keep line breaks (outlets that draw line breaks as escapes or dots merge the next word into the marker, as they already do after any value). Raw and JSON-escaped line breaks are both read; blanks inside escape sequences or control strings never count as indentation, and a block line holding only a control string counts as content and is masked whole. Forms where the marker is followed by more text on the same line, `=`-separated forms such as `key = \|` or the `=>` arrow, and a header at the very end of the text read exactly as in the previous release. From 0.94.0 the private-key header has one definition, generated from the settings contract: the whole-string check finds it anywhere in the text (as the contract's own write-side check does from its 10.0.0 release), and the key-block cutter builds its header line from the same definition, so a lowercase header or one without a space after BEGIN has its key body masked like the canonical form, while public keys and certificates are left as they were. Since 0.94.0 the credential words also read through YAML node properties (type tags such as `!!str` and anchors such as `&name`) between the colon and the value: the properties and the value after them get one marker each, block scalars and block-sequence items included, and a property with no value after it on the same line reads exactly as before. A list after a credential label (a block sequence, a flow sequence including JSON arrays, or a same-line comma list after a plural label) is judged item by item with the same exemptions as a single value: a list of plain key names after a bare `keys`, or of integers after a bare `tokens`, is left alone, and any value-like item masks every non-neutral item. A dash right after the colon on the same line is a list bullet, not a value. The retry overlay's error text (`error.formatted` on the four error arms) passes through the same outlet, and the label vocabulary is pinned as it stands (members masked, known out-of-family keys such as `passphrase` left as written). From 0.95.0 the credential words also read a value wrapped in round or square brackets after a scheme word or label (placeholder-shaped contents such as `(API_KEY)` or `(your-token)` and bracketed prose stay as written), a JSON-escaped tab between a scheme word or label and its value, a same-line associative-array arrow with a quoted label and a quoted value, and a `token` scheme word with its value after a credential label as one value; a word with no label at all is masked when it is shaped like a random token (length, three character classes, entropy and run structure), while hex hashes, UUID lists, base64 image headers and data URIs, integrity strings, provider object ids and ordinary identifiers stay as written; a known secret prefix glued after an underscore or hyphen inside an identifier or a URL path segment is masked from the prefix on; and a masked value no longer swallows a closing backquote at its end, or a closing round or square bracket at the end of a URL that has no matching opener, while a backquote inside a value never cuts it short; an address after `Bearer` or `Basic` that is itself the value of a credential label (including `Authorization:`) is masked whole, keeping the scheme word, while an address with no label stays as written. Credential masking also covers three shapes that used to print as written: a name glued onto a credential word (PGPASSWORD=, CUSTOMSECRET=, dbPassword:, secretAccessKey=; singular only, so token-count names such as maxTokens or inputTokens and English words such as monkey stay visible), Bearer or Basic glued to a preceding word character (_Bearer, xBearer), and a known secret prefix followed by one unbroken run of six or more letters and digits (sk-abcdefghijk). The new readings only add masks: a value that was already masked whole stays masked whole when it contains one of these shapes. The configured-address display and the upload check read the same label rule, so a glued name used as a query key counts as a credential in both. A frozen copy of a pattern net that a downstream host used to run locally is replayed against the output as a no-regression check. A length cap never splits a redaction marker: the cut backs off to before it on every capped path (plain, position-tracking, paired backslashes, and the shared cap of the field-escaping helper); strings that are not complete markers are cut as before. Since 0.98.0 a value glued to a semicolon, a value wrapped in braces or angle brackets (template placeholders stay readable), a bare `Basic` / `Bearer` assigned to an environment-style credential name, and an ordinary word right after `Bearer` / `Basic` in a credential header are masked on every outlet, under a 65 KB linear-time bound; the diagnostic-word predicate answers from the same word list the outlet uses. |
 | `scripts/run-engine-agent-absence-projection-test.mjs` | Absent background agents: when the engine stops reporting a background agent and no final state has arrived, the row is marked absent and this package owns every decision about it, so all clients agree. One predicate says whether a row is absent (the mark, not the status, decides). An absent row keeps its last known status, never counts as running, and is never counted as completed, failed or stopped; its elapsed time stops at the last moment it was seen, and its sentence says it may still be running. The end-of-turn sweep never settles an absent row (or a resident one). A row that comes back, or a real final state for the current cycle, clears the mark; a late final state from an earlier cycle does not. Absent rows are never removed at the short grace window. After the hard limit (30 minutes from the last time they were seen) the host is asked for the background-agent registry reading of each row: only a reading that the agent has ended or is not listed lets the row go, and each removal is returned as a fact the host must act on and announce; a reading of running, unknown, missing or unrecognised keeps the row and schedules nothing, so no standing poll is created. Until a registry reading is available every absent row stays. A row someone is viewing is held and reported separately only once the registry confirms it is gone. The row sentence, the removal sentence and the late-result sentence come from one place, never state an outcome or that the agent finished, and escape control characters in names, in the engine's removal word and in the late-result status. An end-to-end cell drives the real fleet projection and the real absence channel through every decision. From 0.84.0 the package also reads the registry itself: a reader lists the session's background registry only when the server advertises the listing, classifies its failures as unavailable (a failed read keeps the HTTP status and error code when the error carries them, and never invents either), and refuses a partly readable listing as a whole; a classifier turns the listing into the per-row reading, treating a registry status it cannot place as unknown and reading "not listed" only for keys shaped like registry handles and only when the host states the engine is a single process and names the row's session, because a key from another identity space proves nothing by being absent and, on a multi-replica deployment, a listing answered by another replica does not contain this session's agents at all (without that statement a missing row reads unknown and the row stays); and a predicate returns the agents the registry still counts as live that the host has no row for. A listing at the server's 500-row cap cannot show that an agent is gone either, so a missing row there reads unknown; the cap is checked against the engine's own list clamp and, when a server build is supplied, against the server's route. Each listing carries its session and a per-client sequence number, and a listing superseded by a later delivered read, or read for another session than the one the host names, counts for nothing. Only the exact listing object the reader returned can authorize removing a row or adding one: a spread copy, a structured clone, a filtered or a hand-built listing reads unknown and adds nothing, the listing, its row array and every row are frozen, the cap check uses the row count recorded when the page was read, and rewriting a listing's sequence number fools nothing. The fill predicate holds back agents first seen within the absence settle window — timed on one monotonic local clock from when this client's reader first saw the id, so a skew between the server's clock and this one, or reconnecting to a long-running agent, cannot skip the window — agents without a readable registration time and agents the host saw end since the read went out; the registry key of a row, progress or absence event is whichever of its two ids is shaped like a registry handle; and none of the reader, the classifier or the predicate throws. |
@@ -442,13 +442,13 @@ guard still cross-checks the table by name).
 | `scripts/run-gate-interrupt-safety-test.mjs` | The two safety properties of running the suites themselves. The collecting runner no longer kills a suite outright when its time limit expires: it sends a termination signal first, waits a grace period for the suite to clean up, and only then kills it — and it records the suite as timed out however it exits, so a suite that exits cleanly after the signal is still not counted as passing. The limit and the grace period can be widened for a single suite in `gates-manifest.json` (an optional entry with a written reason; a malformed entry, including one with a misspelled field, stops the runner before any suite starts instead of silently falling back to the default, and an entry filed under a misspelled name is reported by this guard rather than ignored), and the negative-control suite is widened there. Every property is exercised on byte-for-byte copies of the real runner and the real negative-control suite in a throwaway directory: a suite that honours the signal finishes within the grace period, one that ignores it is killed when the grace period ends, and the summary lines are unchanged; once a suite has exited the runner waits at most a short drain window for its output pipes, so a child process that inherited them and outlives the suite neither turns a passing suite into a timeout nor holds the runner past the limit, the grace period and that window; the negative-control suite, interrupted in the middle of a rehearsal by any of the three signals or by the runner's own time limit, restores the file byte-for-byte, leaves no backup behind, stops the rehearsed guard together with anything it started, regenerates the build output (checked on a small project in the throwaway directory), and exits with 128 plus the signal number; a rebuild that would overrun the grace period is cut short and its compiler stopped; a backup left behind by an earlier run — next to a later target, loose in the source tree, or inside a symlinked dependency directory — makes it refuse to start without touching anything, naming the file and how to restore it. |
 | `scripts/run-system-reminder-open-tag-test.mjs` | The opening-tag locator for the engine's `<system-reminder>` envelope, for hosts that split a message into envelopes and body text themselves rather than only stripping or unwrapping it. `findSystemReminderOpenTag(text, from?)` returns the start and end of the next opening tag at or after `from` (UTF-16 indices, `end` one past the tag) or `null`, and it recognises exactly the two shapes the engine mints — a bare tag, or a tag with a single `mark` attribute whose value is 22 base64url characters — through the very same matcher the strip and unwrap entry points use, so there is no second grammar to drift. The guard pins both engine shapes to the exact index, refuses the non-engine shapes listed in the integration guide plus near relatives (and does not let them swallow a real tag that follows), refuses an opening tag truncated at the end of the text, answers `null` without throwing for a non-string text and for a `from` outside the integer range 0..length, and shows that the locator ignores block context (a tag inside a block body, an unclosed tag and a nested inner tag are all located — pairing with a close is the caller's loop, as in the strip entry point). On twenty thousand seeded random strings mixing both shapes, truncations, near relatives and nesting, the set of tags the locator finds equals the set derived purely from the observable answers of the strip and unwrap entry points, and a strip and an unwrap rebuilt on the locator agree with the real ones byte for byte. Two timing cells show a single call over many near-miss tags and a full walk with `from` moving forward both stay linear |
 | `scripts/run-detach-durable-off-verdict-test.mjs` | The verdict behind the durable-off hint, now a public entry point without the process-level gate. `isDetachDurableOff400(err)` answers whether an error is the engine's refusal of the detach-on-disconnect header on a deployment with no durable run ledger: status 400 and the engine's own refusal sentence in the message, and nothing else — in particular not the error code that refusal carries, because the engine uses the same generic precondition code for unrelated refusals on the same submit path, and treating those as this one would silently resubmit a turn without the header. Browser and desktop hosts, which never arm the terminal's detach ledger, can now ask the same question instead of keeping their own copy. The guard pins the positive case on the real sentence (with or without the code, with surrounding text, and on the error the SDK actually throws when a stubbed engine answers 400), the negative case on the engine's other refusals that share the code (verbatim, and again through the SDK), non-400 statuses, and eighteen malformed inputs that must answer `false` without throwing. It shows the verdict does not read the process ledger and that the hint still returns nothing before the header has been sent. The verdict never throws: an error object whose `status` or `message` getter throws, a proxy whose trap throws and a revoked proxy all answer `false`, each property is read at most once (and `message` not at all unless the status is 400), and the hint therefore answers nothing for those objects instead of throwing as it did before. Against a frozen copy of the previous hint, every other input gives the same answer byte for byte, armed or not. When the engine's source tree is available, it also checks that exactly one refusal site with that code carries the sentence and that every other one is answered `false` |
-| `scripts/run-parked-resume-startup-test.mjs` | The one reading and the one sentence for a parked resume that fails while starting (`422 parked_resume.startup_failed`). The engine has two unrelated reasons for it — the parked agent's session is gone or the engine rejected the decision itself, or the approval is for a tool the agent inherits from the task that started it and this server cannot hand that tool over — and the response carries **no field that tells them apart**; the reason lives only in the engine's prose, which this package does not branch on. So the reading narrows the cause only by what the caller itself sent: a deny can only be the first reason (the inherited-tool refusal is minted for approvals alone, and the guard pins that upstream premise), while an approval, or a caller that does not say, gets one sentence that lays out both ways forward instead of guessing. Either way the verdict is fixed: sending the same decision again is not a recovery path, and deny stays available. The two sentences are minted in one place, never claim the card is still waiting (in one of the two shapes it is not), and avoid every word the package's own "already decided" scan looks for, since they are appended to failure text that scan reads; all five decision exits append exactly that sentence, and any other failure keeps its text byte for byte. The code also takes precedence over the package's "already decided" word scan: the engine's own text for the gone-session shape contains "not found", which that scan used to read as "someone already settled this card" and answer with a silent reconnect; now a failure carrying this code is classified by the code (no reconnect, the sentence reaches the failure text), while failures with no code or another code are scanned exactly as before. The predicate that says which codes classify themselves is public, so a client with its own "already decided" chain can ask it before scanning words. The engine may also name the reason as a word of its own (an open string; it rides the 422 body's `cause` key, and the SDK reads it as `startupCause` from 13.0.0): the reading carries that word exactly as sent, taking the SDK's named slot first and the generic extra-keys slot second (the only place it exists on SDK 12), only from own data properties and never from the platform's standard error `cause`; an empty, non-string or missing value leaves the key out, and any input without a word reads byte for byte as before. The word changes neither the verdict nor the sentences: the set of words belongs to the engine and has not been published, so nothing here guesses at it (servers up to 7.117.0 do not send it yet). The guard checks every SDK generation it can reach against the real error mapping, and compares a 14,000-input pool without a word byte for byte with the previous release's reading and its two sentences word for word. It also pins the premise that none of the servers and engines it can reach mints a cause word today, so the day one does the guard goes red. |
+| `scripts/run-parked-resume-startup-test.mjs` | The one reading and the one sentence for a parked resume that fails while starting (`422 parked_resume.startup_failed`). The engine has two unrelated reasons for it — the parked agent's session is gone or the engine rejected the decision itself, or the approval is for a tool the agent inherits from the task that started it and this server cannot hand that tool over — and the response carries **no field that tells them apart**; the reason lives only in the engine's prose, which this package does not branch on. So the reading narrows the cause only by what the caller itself sent: a deny can only be the first reason (the inherited-tool refusal is minted for approvals alone, and the guard pins that upstream premise), while an approval, or a caller that does not say, gets one sentence that lays out both ways forward instead of guessing. Either way the verdict is fixed: sending the same decision again is not a recovery path, and deny stays available. The two sentences are minted in one place, never claim the card is still waiting (in one of the two shapes it is not), and avoid every word the package's own "already decided" scan looks for, since they are appended to failure text that scan reads; all five decision exits append exactly that sentence, and any other failure keeps its text byte for byte. The code also takes precedence over the package's "already decided" word scan: the engine's own text for the gone-session shape contains "not found", which that scan used to read as "someone already settled this card" and answer with a silent reconnect; now a failure carrying this code is classified by the code (no reconnect, the sentence reaches the failure text), while failures with no code or another code are scanned exactly as before. The predicate that says which codes classify themselves is public, so a client with its own "already decided" chain can ask it before scanning words. The engine may also name the reason as a word of its own (an open string; it rides the 422 body's `cause` key, and the SDK reads it as `startupCause` from 13.0.0): the reading carries that word exactly as sent, taking the SDK's named slot first and the generic extra-keys slot second (the only place it exists on SDK 12), only from own data properties and never from the platform's standard error `cause`; an empty, non-string or missing value leaves the key out, and any input without a word reads byte for byte as before. The word changes neither the verdict nor the sentences: the set of words belongs to the engine and has not been published, so nothing here guesses at it (servers up to 7.117.0 do not send it yet). The guard checks every SDK generation it can reach against the real error mapping, and compares a 14,000-input pool without a word byte for byte with the previous release's reading and its two sentences word for word. It also pins the premise that none of the servers and engines it can reach mints a cause word today, so the day one does the guard goes red. Since 0.99.0 the start-up cause also accepts the cause words the engine already mints ahead of the SDK vocabulary: the registered ahead words must equal the engine set minus the SDK set in both directions, each is carried through with its own sentence, and a compile-time pin breaks the build when either vocabulary moves without the registration. |
 | `scripts/run-plan-review-injected-wire-test.mjs` | The plan-review orchestration accepts a host-supplied engine client, so a browser host behind a same-origin relay — which cannot install a token-bearing engine target — runs the package's own decision path instead of rebuilding it. With a client injected, the decision, its in-flight latch, the resend-once-without-the-key handling of a `request.field_conflict` refusal, the post-decide status re-pull, the six-state effect classification and the outcome text are byte-for-byte what the installed-target path produces, proven against a real relay-form SDK client and a fake engine while the installed target points at a different fake engine that must receive nothing. The three-choice card depends only on the host's declaration: no engine-version evidence is read, so the cache key a host may still pass is accepted and ignored; every request of the relay-form client leaves without an `authorization` header, checked beside a token-bearing client on the same spy so the absence is not the spy's blindness; and a planted token never reaches a log line, an outcome sentence, a queue item or the host callback on any of the failure paths. A decision may carry a `reason` (trimmed, dropped when blank, capped at the server's limit), an `onOutcome` callback hands the outcome back to hosts that have no model-prompt queue (called once per admitted decision, never for a latched duplicate, and its own failure never affects delivery), a reopen with an injected client no longer needs a host delivery function on a non-default session, and the two capability probes take the same injected client with a bounded wait. A decision that the injected client's own time limit cuts off is reported as sent without an answer within the time limit (it may have taken effect), in the same sentence the installed-target path uses when its limit is reached, while a connection that cannot be made is still reported as unreachable and one that drops after the decision went out is reported as possibly applied (the same no-answer judgement the background-agent verbs use); a client built with the documented decision budget answers a slow approval exactly as the installed target does, and a client whose namespaces are callable functions is accepted as before. The package's own reads through an injected client — the two probes and the post-decide re-pull — settle at their deadlines even while the client sleeps through a retry back-off, and abort the request underneath; a malformed injected connection is reported as not sent, never thrown. |
 | `scripts/run-session-policy-refused-removal-test.mjs` | The way out for a session that already carries a tool name the engine refuses. The engine refuses to start a run when any rule applied to it names a tool its own tool-name check does not call current — a retired tool, a name containing "__" without a protocol prefix, or (engines from 7.48.0) a protocol name whose server or peer segment it never mints — so a session whose own rule record holds such a name fails at startup on every run. The guard pins three pieces. First, the refusal test itself: since 0.97.0 it is the engine's own tool-name check read through `ruleToolNameVerdictOf`, compared name by name, in both directions, with that check called directly over a corpus that includes padded, wildcard and parenthesised forms, and it is shown to be wider than the verdict used before a write — a padded or wildcard form is withheld there for another reason yet still refuses the run, so a removal driven by that verdict would leave the session broken. Every corpus name is then written alone into the deny list and into the allow list of a real engine run: the test says "refused" exactly when the run fails at startup with the refusal code and the model is never called, while the same shapes in the command lists are not audited at all. Second, the removal verb: it takes only the refused entries out of those two lists and leaves every other entry, list and ordering byte-for-byte, keeping an emptied allow list as a present empty list; after it runs, the same session's next real engine run gets past startup, for every refused name in the corpus and in both lists. Against the installed engine as an ordinary caller, taking a deny entry out is refused as a loosening — reported as such, written exactly once, the record unchanged and the next run still refused, never a false success — while an allow-list-only removal, which narrows, succeeds; against a model of the newer contract the engine has announced, the removal succeeds, and the same model still refuses a removal of anything else. A concurrent writer causes exactly one re-read, judged again on the other writer's record, and a second collision is reported rather than retried; read failures, coded write refusals and unconfirmed writes (no verdict, a receipt that cannot be read, or a receipt that does not account for the removal, including one that still holds a removed entry) each land in their own arm, and running the verb again after an unconfirmed write reports nothing left to remove. Third, the wording: one sentence per outcome; the loosening refusal names operators and says the record is unchanged, the unconfirmed one says the change may already be in effect, and none carries a rule string or engine text. The startup-failure sentence answers to that one code only, lists every place the entry can be set without naming who may remove it, and a real run with the name in the caller's own policy rather than in the session record yields the same code — which is why the sentence may not claim the entry is in the session record. The verb also requires the tool roster reported by a run on the deployment: a name listed there, as a tool name or an alias and compared exactly as the engine compares it, is left in place even when it is a retired name, because the engine accepts that rule — with a real engine run that mounts a deployment tool under a retired name (or with that name as an alias), the removal keeps the deny rule and the next run still starts. Without a roster, or with one that cannot be read, nothing is read or written and the outcome says why; a readable empty roster is still a roster. The removal sentence speaks separately about names taken out of the deny list and out of the allow list. |
 | `scripts/run-fixture-flat-done-ratchet-test.mjs` | Test fixtures that feed the engine's terminal `done` frame are kept in the shape the live engine actually sends. Since the terminal record became a tagged cause (`result.terminal`), the older flat shape (`result.status` plus loose keys) reaches a client only in two ways: a stored row replayed verbatim from before the upgrade, and the server's own refusal envelope — so a fixture written in the flat shape tests the replay path while claiming to test the live one. The suite finds every `{ type: 'done', result }` literal under `scripts/`, follows `result` to the object literal it really comes from (in place, through a variable, through a helper's parameter at each of its call sites, through a spread, or through the rows of an iterated array) and counts the flat ones that carry no one-line note saying they model a replayed row or a refusal envelope. That count may only go down: it is a ceiling kept in the ratchet registry, and a count under the ceiling prints a step-down line instead of passing in silence. A planted corpus with a known number of flat, noted and cause-shaped frames in each of those forms must be counted exactly, and a note that sits inside a string or gives no reason does not count |
 | `scripts/run-authority-envelope-mirror-test.mjs` | The authority-envelope tags that a peer's message body is defused against before it is written into a transcript line. Tags the engine uses to speak with its own authority (reminders, completion notices and the like) must never survive inside a body a peer wrote, or a forged completion notice could be read back on resume as a real one and poison the dedup ledger. The package's list is generated at build time from the engine package's public entry point, and the guard checks that the source binds that generated list rather than a hand-written copy and that it matches the installed engine package in both directions: a tag the engine treats as authority and the package lacks is red, and so is a tag the package defuses that the engine does not, since that rewrites ordinary text in a peer's message. It also confirms the published list is the engine module's own value and is still derived from the engine's envelope registry, which it reads from the engine's module by path because the registry itself is not published. At the rendered output, every engine authority tag placed in a body (opening, closing, with attributes, upper case) comes out defused on all three peer lanes, and the engine's non-authority envelope tags come out untouched |
-| `scripts/run-upstream-tables-test.mjs` | Engine facts the package may not import at run time (the notice code register and its audience table, the reasons an injected MCP server is dropped, the protocol namespace prefixes and the retired tool names with their current names) are generated at build time from the installed engine package's own public entry point into plain literal modules, which are committed. The generated files must be byte-identical to what the installed engine produces today, no other file may sit beside them, each names the generator and the engine version it came from, none imports anything, and the package's public names are the generated objects themselves. The generator refuses rather than guesses: a table that is missing or empty, a notice code without an audience row (or the reverse), a third audience value, or a retirement note in a shape it does not know all stop generation instead of producing an empty or partial table, and it reads only the engine package installed in this package's own dependency folder. Two word lists the SDK already exports as values (who asked a question, and which word a mandated approval stands on) are now the SDK's own arrays rather than copies; because those arrays are not frozen, the package's own decisions are shown not to change when they are modified in place. Inside the package, the thinking-level and permission-mode vocabularies, the tier words, the rejection sentence and the cancelled code each exist exactly once in the source. A second batch covers engine fact tables that newer engine packages publish from their entry point (the notice cause and source lists, the run-ending reasons, the reserved control-verb names, the workflow size caps, the authority envelope tags, the denial layers, the reasons a requested permission mode did not take effect, and the structured tool-card types): each is generated the same way and bound where it is used, no hand-written copy of any generated table may remain in the source, and an engine package too old to publish them makes the generator stop and name every missing export rather than fall back to reading private files. A third batch covers the closed set of codes the engine refuses a declared caller tool with: it is generated the same way and folded into the package's configuration-refusal recognition set, which must contain every one of those codes; the set only recognises them (no code path branches on them), an unexpected code shape stops generation, and an engine package that does not yet publish the set makes the generator stop and name it. A fourth batch (from 0.88.0) covers the closed set of reasons a permission mode's release of a question may carry: it is generated the same way, the package's public word list is that generated object, no hand-written copy of its words remains in the source, and an engine package that does not yet publish it makes the generator stop and name it. A sixth batch (from 0.90.0) covers the table of tool-name aliases that hook matchers resolve (the alias spelling of a built-in tool to its canonical name): the engine publishes no ready-made table, so the generator expands the aliases declared on each row of the engine's static tool catalog, and the package reads the result through two public functions (the table itself is not exported). Those functions must agree name by name with the engine's own name resolution, leave every name outside the table untouched (empty strings and inherited object keys included), never fold case or whitespace, and never throw. The generator stops rather than writing an empty table when the catalog declares no alias at all, when a catalog row has an unexpected shape, when an alias collides with another row's canonical name, or when an alias is an integer-like string (object keys of that shape would be reordered), and a run of it against a catalog without aliases exits non-zero and leaves every generated file byte-identical; the generated table is built from key-value pairs and is executed in the check against the engine's own resolver, so an alias spelled like an inherited object key still lands as an own entry The generator also produces five closed sets that the engine exports from its package root — the reasons a rule's evidence can be absent in an approval question, the lanes of the system injection queue, the full list of built-in read-deny tiers, the tiers active by default, and the per-class arbitration switches — as frozen literal tables that the package's own root entry hands out as the very same objects. The check compares them item by item and in order with the engine (the arbitration rows bit by bit, with the key set compared to the engine's own list of ask classes, and the tier lists with the engine's own tier resolver), refuses to generate when a name is missing, a list is empty or has duplicates or non-strings, a default tier is not in the full list, or an arbitration row has a missing, extra or non-boolean switch, and a run of the generator against an engine without one of them exits non-zero and leaves every generated file byte-identical. The reader that answers whether a standing answer may settle an ask reads only the generated arbitration table: its answers follow a table with one switch flipped, a class added or a class removed, the source holds no class name literal, and it treats a throwing getter, a prototype-only field or an unknown class as unknown rather than as permission. The engine code string for a self-contradictory request body appears as a string literal exactly once in the source; the three refusal readers that use it keep their own criteria. From 0.92.0 two more batches come from the engine's package root — four closed sets that stay inside the build, and nine facts the package used to copy by hand (settlement words, classifier refusal causes, the auto-mode unavailable causes, tier order and the familiar tier aliases, the background agent handle shape, the sender permission classes, MCP liveness words and the three rule behaviors) — and a third source is read: the settings contract package, from which the engine-owned hook events, the hook events whose matcher reads a tool name, the known-secret literal shape, the tier group name shape, the budget keys grouped by value type and the model entry interface words are generated into their own file. Each former copy gets a check that the public name is the generated object itself, or, for module-private constants, that the initialiser is the imported generated identifier; regex and alias-table copies are scanned for alongside array copies; and malformed upstream shapes, extra members and missing names each make the generator refuse. Two numeric limits are generated as well (the largest inline image in base64 characters, and the longest follow-up text): the same value as the engine's export, never zero, absent or malformed, not on the root entry, read by the human-submission assembly through the generated constants with no hand-written copy left there. The closed set of causes by which a deployment hook can keep a prompt from being submitted is generated as well: the same words in the same order as the engine's export, frozen, a literal tuple type, exported at the root as the generated object itself, agreed with by the engine's own membership test, and with no copy of its words in the reader that uses it; an extra member changes the generated output, while a malformed shape or a missing name makes the generator refuse. Since 0.94.0 the generator also derives the request-side permission-mode words and the user-input lane table (lane to notice code and count key); the package-side fold of the arbitration table takes an optional question face, reads the releasable classes per face from the generated ask-face tables only, and is diffed against the engine's own fold on every face. From 0.94.0 the settings-contract batch also generates the private-key header definition, which must be unanchored and judges the same samples as the contract's own check; a missing, anchored or stateful definition makes the generator refuse, and the display module's header constant is the generated object itself. Since 0.94.0 it also generates the ask-face tables (the face words, the tool names on the shell face derived from the engine's own face function over its catalog, and the releasable classes per face), checks them against the engine item by item and against the arbitration table, and refuses to generate on every malformed shape or missing export. From 0.95.0 a settings-contract function batch copies three pure functions from the contract package's entry point (the narrow loopback check used before sending no credentials, the hierarchical URL check, and the @-mention narrowing) by their own source text: the generator refuses when a function reaches for any name outside itself except an allowed global, the copies must answer exactly like the originals on independent samples, and the package's readers use the copies instead of their own versions. The configuration-refusal registry has no read point inside the package (the only reading of "is this a configuration refusal" is the prefix predicate), and its own comment says so. Since 0.96.0 the approval mandate word list is generated from the engine package root too, followed by the one word older supported engines still send. Since 0.96.0 the settings contract's hook-matcher reading (`hookMatcherSelects` / `hookMatcherForm`, with the two private declarations they depend on copied verbatim from the installed module) is generated the same way and must answer exactly as the installed contract on an independent sample set; generation is refused when the module touches those private declarations anywhere else (modified at load time, reassigned, or exported), because then the copied text would not be what runs upstream. |
-| `scripts/run-retirement-ledger-test.mjs` | Transitional code whose retirement condition used to live only in a comment now has a machine-readable trigger. `scripts/retirement-ledger.json` lists each piece (a compatibility read, a mirrored table, a normalizer inside the differential check) with an anchor into its own code, the retirement condition quoted from the source comment, and one of three trigger kinds: this package's version reaching a retire-by version; the installed upstream package reaching a version, exporting a name from its entry, or declaring a member on a given type (read with the TypeScript checker); this package's declared support floor (`scripts/support-floor.json`) for the server, the engine or the SDK reaching a given version (an unreadable or malformed declaration fails the check instead of being skipped); or the reference client tree used by the differential check having moved to a given version of this package. The check fails when a trigger has fired and the transitional code is still there, when a row's anchor can no longer be found (the code is gone but the row stayed), when the quoted condition no longer matches the source comment, when a row has no machine trigger at all, when a retire-by version sits more than three minor lines ahead, and when the anchor of an already retired piece reappears. A reading that cannot be taken is reported as skipped when the absence is legitimate (no reference tree on disk, an upstream declared only as a peer and not installed) and is a harness fault otherwise. Before giving any verdict the judge proves itself on in-memory fixtures and on one known-present and one known-absent name in the installed upstream types. Rows already moved to the retired list may also carry the names, keys and function arities that the retirement removed; for those rows the check confirms the names are absent from the package root entry (as exports and as runtime keys), from the built output, and from every check script's code positions (comments excluded), that each retired function has its reduced parameter count (taken from every type signature as well as the runtime length, so an added optional, default or rest parameter is seen), and that a compile-time pin on the root entry's keys holds in both directions — no retired name present, every known live name still present, with a planted live name that must fail. Rows may carry a structured blockedOn (upstream ticket + landing criterion + a ruling id cited in the ruling text): a due row with it prints as BLOCKED-DUE instead of failing, capped at 3 such rows. |
+| `scripts/run-upstream-tables-test.mjs` | Engine facts the package may not import at run time (the notice code register and its audience table, the reasons an injected MCP server is dropped, the protocol namespace prefixes and the retired tool names with their current names) are generated at build time from the installed engine package's own public entry point into plain literal modules, which are committed. The generated files must be byte-identical to what the installed engine produces today, no other file may sit beside them, each names the generator and the engine version it came from, none imports anything, and the package's public names are the generated objects themselves. The generator refuses rather than guesses: a table that is missing or empty, a notice code without an audience row (or the reverse), a third audience value, or a retirement note in a shape it does not know all stop generation instead of producing an empty or partial table, and it reads only the engine package installed in this package's own dependency folder. Two word lists the SDK already exports as values (who asked a question, and which word a mandated approval stands on) are now the SDK's own arrays rather than copies; because those arrays are not frozen, the package's own decisions are shown not to change when they are modified in place. Inside the package, the thinking-level and permission-mode vocabularies, the tier words, the rejection sentence and the cancelled code each exist exactly once in the source. A second batch covers engine fact tables that newer engine packages publish from their entry point (the notice cause and source lists, the run-ending reasons, the reserved control-verb names, the workflow size caps, the authority envelope tags, the denial layers, the reasons a requested permission mode did not take effect, and the structured tool-card types): each is generated the same way and bound where it is used, no hand-written copy of any generated table may remain in the source, and an engine package too old to publish them makes the generator stop and name every missing export rather than fall back to reading private files. A third batch covers the closed set of codes the engine refuses a declared caller tool with: it is generated the same way and folded into the package's configuration-refusal recognition set, which must contain every one of those codes; the set only recognises them (no code path branches on them), an unexpected code shape stops generation, and an engine package that does not yet publish the set makes the generator stop and name it. A fourth batch (from 0.88.0) covers the closed set of reasons a permission mode's release of a question may carry: it is generated the same way, the package's public word list is that generated object, no hand-written copy of its words remains in the source, and an engine package that does not yet publish it makes the generator stop and name it. A sixth batch (from 0.90.0) covers the table of tool-name aliases that hook matchers resolve (the alias spelling of a built-in tool to its canonical name): the engine publishes no ready-made table, so the generator expands the aliases declared on each row of the engine's static tool catalog, and the package reads the result through two public functions (the table itself is not exported). Those functions must agree name by name with the engine's own name resolution, leave every name outside the table untouched (empty strings and inherited object keys included), never fold case or whitespace, and never throw. The generator stops rather than writing an empty table when the catalog declares no alias at all, when a catalog row has an unexpected shape, when an alias collides with another row's canonical name, or when an alias is an integer-like string (object keys of that shape would be reordered), and a run of it against a catalog without aliases exits non-zero and leaves every generated file byte-identical; the generated table is built from key-value pairs and is executed in the check against the engine's own resolver, so an alias spelled like an inherited object key still lands as an own entry The generator also produces five closed sets that the engine exports from its package root — the reasons a rule's evidence can be absent in an approval question, the lanes of the system injection queue, the full list of built-in read-deny tiers, the tiers active by default, and the per-class arbitration switches — as frozen literal tables that the package's own root entry hands out as the very same objects. The check compares them item by item and in order with the engine (the arbitration rows bit by bit, with the key set compared to the engine's own list of ask classes, and the tier lists with the engine's own tier resolver), refuses to generate when a name is missing, a list is empty or has duplicates or non-strings, a default tier is not in the full list, or an arbitration row has a missing, extra or non-boolean switch, and a run of the generator against an engine without one of them exits non-zero and leaves every generated file byte-identical. The reader that answers whether a standing answer may settle an ask reads only the generated arbitration table: its answers follow a table with one switch flipped, a class added or a class removed, the source holds no class name literal, and it treats a throwing getter, a prototype-only field or an unknown class as unknown rather than as permission. The engine code string for a self-contradictory request body appears as a string literal exactly once in the source; the three refusal readers that use it keep their own criteria. From 0.92.0 two more batches come from the engine's package root — four closed sets that stay inside the build, and nine facts the package used to copy by hand (settlement words, classifier refusal causes, the auto-mode unavailable causes, tier order and the familiar tier aliases, the background agent handle shape, the sender permission classes, MCP liveness words and the three rule behaviors) — and a third source is read: the settings contract package, from which the engine-owned hook events, the hook events whose matcher reads a tool name, the known-secret literal shape, the tier group name shape, the budget keys grouped by value type and the model entry interface words are generated into their own file. Each former copy gets a check that the public name is the generated object itself, or, for module-private constants, that the initialiser is the imported generated identifier; regex and alias-table copies are scanned for alongside array copies; and malformed upstream shapes, extra members and missing names each make the generator refuse. Two numeric limits are generated as well (the largest inline image in base64 characters, and the longest follow-up text): the same value as the engine's export, never zero, absent or malformed, not on the root entry, read by the human-submission assembly through the generated constants with no hand-written copy left there. The closed set of causes by which a deployment hook can keep a prompt from being submitted is generated as well: the same words in the same order as the engine's export, frozen, a literal tuple type, exported at the root as the generated object itself, agreed with by the engine's own membership test, and with no copy of its words in the reader that uses it; an extra member changes the generated output, while a malformed shape or a missing name makes the generator refuse. Since 0.94.0 the generator also derives the request-side permission-mode words and the user-input lane table (lane to notice code and count key); the package-side fold of the arbitration table takes an optional question face, reads the releasable classes per face from the generated ask-face tables only, and is diffed against the engine's own fold on every face. From 0.94.0 the settings-contract batch also generates the private-key header definition, which must be unanchored and judges the same samples as the contract's own check; a missing, anchored or stateful definition makes the generator refuse, and the display module's header constant is the generated object itself. Since 0.94.0 it also generates the ask-face tables (the face words, the tool names on the shell face derived from the engine's own face function over its catalog, and the releasable classes per face), checks them against the engine item by item and against the arbitration table, and refuses to generate on every malformed shape or missing export. From 0.95.0 a settings-contract function batch copies three pure functions from the contract package's entry point (the narrow loopback check used before sending no credentials, the hierarchical URL check, and the @-mention narrowing) by their own source text: the generator refuses when a function reaches for any name outside itself except an allowed global, the copies must answer exactly like the originals on independent samples, and the package's readers use the copies instead of their own versions. The configuration-refusal registry has no read point inside the package (the only reading of "is this a configuration refusal" is the prefix predicate), and its own comment says so. Since 0.96.0 the approval mandate word list is generated from the engine package root too, followed by the one word older supported engines still send. Since 0.96.0 the settings contract's hook-matcher reading (`hookMatcherSelects` / `hookMatcherForm`, with the two private declarations they depend on copied verbatim from the installed module) is generated the same way and must answer exactly as the installed contract on an independent sample set; generation is refused when the module touches those private declarations anywhere else (modified at load time, reassigned, or exported), because then the copied text would not be what runs upstream. Since 0.99.0 it also covers three more generated groups: the request-attachment seat description, its three closed sets and two limits match the installed engine entry for entry, stay frozen and off the root entry, and have no readers outside the generated directory; the interface-language value set is generated and the public UI_LANGUAGES is that same array; and the hooks-configuration classifier is copied with its private closure into its own generated file (the only file in src exempt from type checking), bound to the generated hook-event table and re-exported with the settings contract's own declared signature, answering exactly like the installed original on two independent sample sets. The types in that signature, with everything they depend on, are copied declaration by declaration from the contract's declaration files into a generated type mirror, so no source file and no shipped declaration file references the development-only contract package; a compile check inside the guard proves every mirrored type and the installed one are assignable to each other in both directions, and two planted edits, each breaking one direction only, are shown to fail in exactly that direction. The generated files also cover the contract package's request-plane MCP entry judges (seven exports plus the helpers they depend on, copied as one closure whose two cross-module dependencies are factory parameters bound to the generated word list and prefix guard) and the engine's ordered list of ask classes, which the ask-class pair reader now reads instead of borrowing the arbitration table's key order; independent closure derivation, byte-for-byte declarations, same-answer pools, an empty-context binding witness and refusal controls guard them. |
+| `scripts/run-retirement-ledger-test.mjs` | Transitional code whose retirement condition used to live only in a comment now has a machine-readable trigger. `scripts/retirement-ledger.json` lists each piece (a compatibility read, a mirrored table, a normalizer inside the differential check) with an anchor into its own code, the retirement condition quoted from the source comment, and one of three trigger kinds: this package's version reaching a retire-by version; the installed upstream package reaching a version, exporting a name from its entry, or declaring a member on a given type (read with the TypeScript checker); this package's declared support floor (`scripts/support-floor.json`) for the server, the engine or the SDK reaching a given version (an unreadable or malformed declaration fails the check instead of being skipped); or the reference client tree used by the differential check having moved to a given version of this package. The check fails when a trigger has fired and the transitional code is still there, when a row's anchor can no longer be found (the code is gone but the row stayed), when the quoted condition no longer matches the source comment, when a row has no machine trigger at all, when a retire-by version sits more than three minor lines ahead, and when the anchor of an already retired piece reappears. A reading that cannot be taken is reported as skipped when the absence is legitimate (no reference tree on disk, an upstream declared only as a peer and not installed) and is a harness fault otherwise. Before giving any verdict the judge proves itself on in-memory fixtures and on one known-present and one known-absent name in the installed upstream types. Rows already moved to the retired list may also carry the names, keys and function arities that the retirement removed; for those rows the check confirms the names are absent from the package root entry (as exports and as runtime keys), from the built output, and from every check script's code positions (comments excluded), that each retired function has its reduced parameter count (taken from every type signature as well as the runtime length, so an added optional, default or rest parameter is seen), and that a compile-time pin on the root entry's keys holds in both directions — no retired name present, every known live name still present, with a planted live name that must fail. Rows may carry a structured blockedOn (upstream ticket + landing criterion + a ruling id cited in the ruling text): a due row with it prints as BLOCKED-DUE instead of failing, capped at 3 such rows. A blockedOn may name a calendar date instead of a ticket (for a condition that is itself a date): a due row then prints as BLOCKED-DUE only while today is before that date, and fails once the date has passed or today cannot be read. |
 | `scripts/run-plan-review-decision-ledger-test.mjs` | The plan-review reopen path refuses on one decision ledger shared by every delivery path, not only the package's own POST window: a decision the host reports as handed over (before the card's answer reaches the package, or on the host's own delivery path) counts as in flight until it settles; a pending-row read that started before a decision settled is a stale snapshot of the gate just answered, so a reopen carrying that read mark refuses; and a decision proven not to have taken effect (nothing sent, a refusal that by contract changed nothing, or the engine proven to still hold the very same gate instance — a status alone does not prove it, since a run that moved on to a new plan gate reports the same status) neither counts as in flight nor makes such a snapshot stale, so a failed decision does not consume a legitimate reopen. The ledger's capacity bound never evicts the package's own in-flight decisions, so a duplicate approval stays latched however many decisions are in flight. The package's own in-flight window now runs from the POST to the moment the outcome is emitted (settlement happens before the outcome is queued, so a read the host starts from the queue port is not stale), while the duplicate-delivery latch keeps its POST-only window. A handed-over decision the package's delivery path then sends is the same ledger entry (the host's early settle does not release it); two decisions on one run each settle on their own; positive evidence that the run left the gate settles what is open without releasing the latch; bad inputs neither throw nor record. The ledger is checked against the terminal's current implementation vector by vector when that tree is on disk. Positive evidence that the run left the gate can arrive after the host has handed a decision over and before the package's delivery path picks it up (while the host is still waiting on its gate snapshot, or in the microtasks between that wait ending and the delivery call); the delivery path then still picks up that same entry instead of minting a new one, so the latest decision mark the host keyed its snapshot and its not-applied notice on does not move, the duplicate-delivery latch and the capacity rule hold for it, and the three failure outcomes read byte for byte as they would without the hand-over. Only an entry settled by such evidence is picked up again: one the host settled itself, one it revoked, and one that already produced an outcome are not, and a new decision gets a new mark |
 | `scripts/run-subagent-injected-wire-test.mjs` | The background-agent verbs — the child report read, the live tail, the task-handle output read and stop, steer, the resume assembly, manual compaction with its capability warm-up, and the delegated-prompt read — accept a host-supplied engine client, and the plan-review orchestration now resolves its client through the same single construction point, so a browser host behind a same-origin relay runs every one of these paths without an installed engine target. With a client injected, each entry sends only through it: proven against a real relay-form SDK client and a fake engine while the installed target (default and keyed slots alike) points at a different fake engine that must receive nothing, with the row's own session parameter carried, no `authorization` header on any request (checked beside a token-bearing client on the same spy), and a planted token never reaching a log line, a return value or a failure detail on the injected, installed-target and unreachable paths. The capability evidence behind the task-handle verbs and the row stop gate is read only under the key the host names, never from the installed target; the report, tail and compaction warm-up cache their own probe under that key and re-probe when it is absent. A malformed injected client or connection lands on each entry's existing cannot-send outcome (`null`, `no-wire`, `unavailable`, `offline`, `unreadable`) with zero requests, never a throw or an unhandled rejection. Every request through an injected client settles at the package's own deadline even while the client sleeps through a retry back-off and aborts the request underneath; the caller's own signal still applies, and the old signatures answer as before apart from the no-answer outcomes described next. A steer, resume or stop that may have reached the engine but got no answer — the package's own deadline on an injected client, the client's per-request limit on either path, or a transport failure that cannot be shown to have happened before anything was sent (the connection dropped after the request went out) — is reported with a machine-readable `unconfirmed: true` beside the unchanged `reason`, so a host can say it may have taken effect instead of calling it a failure; a cancellation by the caller after the request went out is reported the same way (only the wait was cancelled), while a real 4xx or 5xx answer, a connection that could not be made at all, a verb that threw before returning a promise, and a cancellation already in place before the call (nothing is sent) carry no such key. An installed target is copied by value when a call resolves it, so a delayed or retried compaction still goes to the engine and credentials it was resolved against even if the host's target object changes meanwhile, and a verb that throws synchronously is classified exactly like one that rejects. The resume failure classifier that hosts reuse for their own direct calls makes the same judgement from the same single source, so every client renders it the same way. A successful sub-agent steer carries the server's acceptance id (`messageId`) only when the receipt holds a non-empty string; older servers leave it out. |
 | `scripts/run-session-background-stop-test.mjs` | The session-level stop for background work (`stopEngineSessionBackground`): one call asks the engine to stop the background tasks of a session, with `includeRetained` passed through as the required true / false choice. It resolves its client through the same single construction point as the other engine verbs (a host-supplied client, or the installed engine target for the chosen session slot) and sends nothing unless the engine has declared the stop face (`background.exitFaces` strictly `true`) under the capability key that applies: evidence that is missing or unread, an older engine without the face, a malformed flag, a flag found only on the prototype chain, or evidence recorded for a different deployment all end in `unavailable` with zero requests. The request body on the wire is exactly `{"includeRetained": <boolean>}` whatever object the caller passes. A 200 answer returns the receipts verbatim (identifier and outcome word; a row whose word it does not recognise is kept with that word as it is, so the host can still name it and the companion below sorts it as may still be running), counts rows whose identifier or word it cannot read instead of inventing them, and leaves the receipt list absent rather than empty when the body cannot be read. Refusals are sorted by machine code before HTTP status (missing service credential, no ownership face, unknown session versus missing route on the same 404, request shape, not permitted, rate limited with the server's wait hint, anything else), a server failure is an `error`, and each call sends exactly one request even through a client configured to retry. A request that may have reached the engine but got no answer — a time limit, a connection that dropped after the request went out, or a cancellation by the caller after the request went out — carries `unconfirmed: true`, while a real answer, a connection that never opened, or a cancellation already in place before the call (nothing is sent) does not. Malformed arguments, options, connections, clients, capability records, answer bodies and error objects never throw and never leave an unhandled rejection, and a planted token never reaches a log line or a result. A pure companion (`engineSessionBackgroundReceiptReadingOf`, with the frozen state list `ENGINE_SESSION_BACKGROUND_RECEIPT_STATES` derived from its wording table) sorts each receipt word into stopped, still running (kept on purpose) or may still be running, with one fixed sentence per state so every client says the same thing: each of the four words lands in its state, while any word it does not recognise — a future word, a different letter case, stray whitespace, a near spelling, a prototype-chain name — lands in may still be running with the word carried back verbatim, and non-string input (boxed strings and objects with their own `toString` included) lands there too without being coerced and without throwing; it never reports stopped for anything it cannot confirm. The three sentences differ, the two that are not stopped never use the word stop, and the state list and the states actually produced match in both directions. |
@@ -468,7 +468,7 @@ guard still cross-checks the table by name).
 | `scripts/run-prompt-delivery-projection-test.mjs` | Whether a user message actually reached the engine, judged once for every client. A ledger records, per message id and per submitted batch, that a submission entered the submit path and whether the engine created a run for it: a stream frame from the new run counts, a later accepted steer into the running run counts, while the session-busy refusal does not, and neither do the frames a server writes before creating the run (the stream-head `meta` frame and replays of approval cards still pending from earlier runs), which arrive ahead of that refusal. Both sets only grow, so the answers do not depend on arrival order, replays or repeats; a message never seen on the submit path is reported as unknown rather than undelivered. Two row-level readers filter side-channel payloads (only messages known to be undelivered are removed, everything else is kept in order). Records serialize one per line in the shape hosts already persist. The check also replays the exact frames an installed server package writes for a busy session through the real SDK stream parser when one is available and reports that part as skipped otherwise. The idempotent-replay form of the busy-session refusal (the raw conflict body without a status, sent when a host reuses an idempotency key while the first submission is still in flight) does not count as delivered either. Since 0.90.0 the check also feeds the delivery frames a steer now produces (a queued frame and an applied frame on one input id, which equals the message id on the receipt): the first records delivery, the second changes nothing, replays and reversed order give the same records, and the answer does not depend on whether the input id equals the message id, differs from it, or is absent. The ledger object itself also carries the two row-level readers as methods bound to it, `knownUndeliveredMessage(message)` and `withoutUndelivered(messages)`, which answer exactly what the two free functions answer for that ledger on every input shape: a bare string is not a message row, so it is judged false, and that is why the row-level method does not reuse the name of the id-keyed `knownUndelivered(uuid)`. They keep answering when taken off the ledger, when the ledger is copied, and after it is frozen, they read the ledger as it is now, and a ledger rebuilt from stored records carries them too. The check also compiles the installed declaration file against positive and negative type examples, including a hand-written nine-member ledger stand-in that must still be accepted by the two free functions (their ledger parameter only needs the id-keyed reader). Since 0.91.0 every delivered record the ledger writes names its source (`by`: a frame from the new run, or the accepted steer receipt) and may carry a share (`full` or `partial`, absent meaning full); a delivered record without a source, as written by earlier versions, is read as unknown rather than delivered, a record whose extension fields cannot be read falls back to that unknown form and is stored without them, and keys the reader does not know are dropped and counted. Merging takes the highest of unknown, partial and full for each message, whatever the order: the check replays 2400 random sequences in original, shuffled, duplicated, split-and-merged and persisted forms against an independent model, and feeds records written by this version to the previous release's reader, which must give the same answers. A submission can also carry the engine generation it was sent on; the ledger then lists, once per generation, the messages submitted there that are still known undelivered, without changing the recorded facts, and a malformed generation is ignored while the submission itself is still recorded. Revoked proxies passed as batches, outcomes, record lists or message lists are treated as unreadable instead of throwing. Both write ports also take the two tiers of an encoded submission, `{ full, partial }` as returned by `deliveredUuidsOf`: the partial tier is recorded with `share: 'partial'`, so a message that was only partly sent is never read back as known undelivered, and an unreadable batch records nothing without throwing. On read-back, once a record's `u` and `s` are read, a throwing extension field or key enumeration keeps the record as legacy instead of dropping it. |
 | `scripts/run-session-memory-erase-projection-test.mjs` | The session owner's memory erase (`eraseSessionMemory`, with the pure judge `sessionMemoryEraseOutcome`): one call sends exactly one request to erase what a session contributed to memory, with a body of exactly the request id plus the unevidenced-erase consent only when the caller gives it as a boolean (never added when absent, and no other key the caller passes reaches the wire); a dot-segment session id is refused locally. The connection takes the same configuration as the engine client factory, and the outbound headers are compared byte for byte with a client that factory builds for every credential form (string, per-request value function, a value function that returns nothing or throws, empty, same-origin relay, loopback without a token, blank or missing principal); a token-bearing connection in a browser host sends nothing. Malformed arguments, options and connections never throw and send nothing. A request that may have reached the engine without a readable answer (time limit, a connection dropped after the request went out or while the body was read, a cancellation after sending, an unreadable response) is reported as no answer, while a connection that never opened, a cancellation before sending or a synchronous fetch failure is reported as not sent. The judge sorts each answer into erased, partly erased, refused (nothing was erased: not sent, or a named engine code issued before any change) or not confirmed (no answer, an unreadable receipt or one for a different request or session, a refusal with two possible meanings, an unrecognised code, or a non-success status without the engine's own error code; a code found only on the prototype chain does not count), by machine code before HTTP status, with one fixed sentence per outcome and a retry posture; a caller who gave the unevidenced consent always gets a person decides on the not confirmed outcomes. Receipts are read with the existing erase receipt reader. The closed lists are frozen, derived from the wording tables and equal to what the judge produces; sentences differ, refusals always say nothing was erased, and none copies a server sentence. The check also boots installed server packages of the current and the previous generation, when available, with their real dispatcher and handler and the engine's real file memory store, and drives the call against them: real receipts and replays, a request id reused on another session, a store without erase evidence with and without consent, unknown and foreign sessions, missing credentials, missing faces, malformed bodies and the coded refusals; the previous generation's untyped internal errors are judged not confirmed. The unevidenced-erase consent counts only when it is the caller's own property (an inherited one is never sent), a session id that cannot be encoded into the path is refused locally without a request and without a rejection, and a signal-shaped object that is not a real abort signal is refused as bad input. |
 | `scripts/run-disable-skills-wire-test.mjs` | The request field that turns off every skill for one request (the deployment's skills, skills the client sends and skills listed on agent definitions). It is carried on the two user lanes only as the literal true; false or an absent value sends nothing, and any other value is refused before anything is sent. A request that also carries a non-empty skill list, or an agent definition with skills, is refused by name both when it is built and again after agent definitions are added later, because the engine rejects that combination. The value reaches a request only through a helper that reads the engine's capability report: when the engine has not reported the capability, reports it as unavailable, has not been read yet, or reports it in a form that cannot be read, nothing is sent (an engine that does not know the field rejects the whole request) and the caller receives the reason plus one sentence saying skills stay on for this run. Requests that do not use the field are byte-identical to the release before it was introduced, apart from images on the one-shot lane, which gained a seat in a later release and are checked by the request-receipt and human-submission guards instead. |
-| `scripts/run-permissions-declaration-test.mjs` | The request-side assembly of permission rules (`permissionsDeclaration`): the deny, ask and allow lists the client resolved from its settings become the `settings.permissions` fragment of a request. Every list is ordered by where each rule came from (managed policy first, then command-line flags, the settings flag, the session, user, local and project settings, and rules of unknown origin last) and filtered by the same judges the engine uses, so a rule the engine would refuse does not turn a whole request into an error — the judges are the client's installed engine functions, so the server can still refuse the whole request when its engine version differs from the client's, when a judge fails on one rule and that deny or ask rule goes through unchecked, or when a list is longer than the engine's limit. A whole-server or whole-peer rule (`mcp__<server>`, `a2a__<peer>`; one rule for every protocol prefix the installed engine knows) is rewritten to the same name followed by `__*`, after checking, protocol by protocol, that the engine judges both forms alike and that every tested server generation accepts the rewritten form. A whole tool name that the engine's rule compiler refuses (for example one containing a closing parenthesis or a backslash) is kept on the client as well, whether or not the rule was rewritten, so every rule sent has passed the same judge; this is checked on generated rule strings and differs from the previous terminal's behaviour in exactly that one registered class. Allow rules are sent only when the engine reports, through one capability position read by `observedPermissionsAllowGrant`, that it reads allow rules as rules rather than as an exclusive list; an engine that does not say so may read an allow list as the only tools permitted, so nothing is sent and approvals keep being answered on the client side. A sent allow rule records that someone answered for the tool; it does not waive the prompts the engine raises on its own, so the client keeps answering matching approvals from all of its allow rules, and allow rules with a content pattern are never sent. Allow rules that come from settings files (or whose source is unknown) are also held back whenever the host does not report that the settings read which produced them was clean (`settingsRead: 'clean'`; a partial or unreadable read, or no report at all, holds them back), while allow rules from the command line or the session are still sent, so every lane follows one rule decided in one place. The capability position is the one servers publish from 7.115.0 on, checked against the capabilities route of each server generation the gate is given. An empty list is never sent, missing rule judges raise an error instead of sending unchecked rules (a judge that is present but fails on a rule is the exception above), the diagnostic sentence is derived from the same judgement as the sending decision and never throws, and inputs with inherited or malformed fields, or rule lists whose gaps are filled from a modified array prototype, are refused rather than silently losing a rule. A companion report, `permissionsDeclarationReport`, comes out of the same single pass: its `permissions` are byte-identical to the sending decision, every rule string lands exactly once either in what is sent (with a table mapping each rewritten form back to the rule that produced it) or in a `kept` list with one reason from a closed vocabulary, and deny or ask rules sent while the rule compiler could not answer are listed as unchecked; the gate keeps one witness per reason, checks that both entry points make the same sequence of compiler calls and throw the same errors, and compares the sending decision with the previous release on two independent rule pools. From 0.92.0 the allow list's whole tool names are checked under the tightening reading, because the engine's rule compiler no longer takes an allow reading at all: the guard proves the swap changes nothing on the wire — the published 0.90.0 build running on the engine generation that receives this list and the current build running on either engine generation emit byte-identical requests over two rule pools, every name the receiving generation refuses under the allow reading is refused by the current compiler under the deny reading and no other, and the mirrored rule shape's behavior set equals the installed compiler's. It also records why the engine's own allow-list grammar is not used as the judge yet: on the names this assembly checks, that grammar and the receiving generation's compiler disagree in exactly five registered classes (names with a backslash, which the receiving generation refuses and would reject the whole request for, are never sent), and a sixth class appearing turns the guard red. From 0.92.0 it also pins how path rules relative to the task root (`src/**`, `./x`, `/x`) are handled: they are sent exactly as written, byte-identical to every published build from 0.90.0 on (0.91.0 included, whose companion report is compared in full as well whenever the capability that turns allow rules on is not reported as true), and never rewritten to an absolute `//` form; for every server generation the gate is given, it records verbatim how that server's settings intake answers such deny and ask rules with and without a task root (refused without one, accepted with one, absolute `//` rules accepted either way), so a change on the server side turns the guard red. From 0.92.0 two of those classes (protocol names whose server segment carries a backslash or a wildcard) are held back before the name is checked at all, so three classes reach that judge, and the guard checks each held witness was never asked. Protocol rule names whose server segment the engine never mints as written (a wildcard segment, or one the engine's own naming function would rewrite) match no tool on any server generation: in the allow list they are held back with the reason `unminted_protocol_segment`, while deny and ask rules are still sent exactly as the previous release sent them (a tightening rule must not quietly disappear; from 7.116.0 the server refuses the whole request and names the rule) and are listed in the report's `unreachable` table with the prefix the engine mints for that server, so a client can warn before sending; the guard compares the deny and ask lists with the previous release byte for byte on a pool full of such names, and checks that the table lists exactly the sent rules the server's judge would refuse. Without the naming function installed the allow copies are held as not judged and only wildcard segments are listed. Since 0.94.0 the allow bucket follows the engine generation that receives it: with the allow-layer capability (and the engine's allow-list reader installed) content-form allow rules are sent as written, and an entry the engine would read as voiding the whole allow list is held back and reported; older engines keep the previous projection byte for byte. The one-sentence notice for the print lane when settings-file allow rules were held back because the settings read was not clean (`permissionsAllowKeptNotice`) is pinned here: the wording per read-health state, the managed-policy variant, no rule text in the sentence, malformed input treated as absent. 0.98.0: the deny / ask buckets now carry every non-empty rule (only equivalent rewrites; the report has no unchecked / unreachable tables; deny / ask kept rows are only empty_rule / duplicate), the allow bucket is byte-identical to the previous build, every throw from the assembly carries a machine-readable refusal identity (read from own data slots, identical in a minified bundle), and the live server fixture refuses / accepts each passed-through rule as expected. |
+| `scripts/run-permissions-declaration-test.mjs` | The request-side assembly of permission rules (`permissionsDeclaration`): the deny, ask and allow lists the client resolved from its settings become the `settings.permissions` fragment of a request. Every list is ordered by where each rule came from (managed policy first, then command-line flags, the settings flag, the session, user, local and project settings, and rules of unknown origin last) and filtered by the same judges the engine uses, so a rule the engine would refuse does not turn a whole request into an error — the judges are the client's installed engine functions, so the server can still refuse the whole request when its engine version differs from the client's, when a judge fails on one rule and that deny or ask rule goes through unchecked, or when a list is longer than the engine's limit. A whole-server or whole-peer rule (`mcp__<server>`, `a2a__<peer>`; one rule for every protocol prefix the installed engine knows) is rewritten to the same name followed by `__*`, after checking, protocol by protocol, that the engine judges both forms alike and that every tested server generation accepts the rewritten form. A whole tool name that the engine's rule compiler refuses (for example one containing a closing parenthesis or a backslash) is kept on the client as well, whether or not the rule was rewritten, so every rule sent has passed the same judge; this is checked on generated rule strings and differs from the previous terminal's behaviour in exactly that one registered class. Allow rules are sent only when the engine reports, through one capability position read by `observedPermissionsAllowGrant`, that it reads allow rules as rules rather than as an exclusive list; an engine that does not say so may read an allow list as the only tools permitted, so nothing is sent and approvals keep being answered on the client side. A sent allow rule records that someone answered for the tool; it does not waive the prompts the engine raises on its own, so the client keeps answering matching approvals from all of its allow rules, and allow rules with a content pattern are never sent. Allow rules that come from settings files (or whose source is unknown) are also held back whenever the host does not report that the settings read which produced them was clean (`settingsRead: 'clean'`; a partial or unreadable read, or no report at all, holds them back), while allow rules from the command line or the session are still sent, so every lane follows one rule decided in one place. The capability position is the one servers publish from 7.115.0 on, checked against the capabilities route of each server generation the gate is given. An empty list is never sent, missing rule judges raise an error instead of sending unchecked rules (a judge that is present but fails on a rule is the exception above), the diagnostic sentence is derived from the same judgement as the sending decision and never throws, and inputs with inherited or malformed fields, or rule lists whose gaps are filled from a modified array prototype, are refused rather than silently losing a rule. A companion report, `permissionsDeclarationReport`, comes out of the same single pass: its `permissions` are byte-identical to the sending decision, every rule string lands exactly once either in what is sent (with a table mapping each rewritten form back to the rule that produced it) or in a `kept` list with one reason from a closed vocabulary, and deny or ask rules sent while the rule compiler could not answer are listed as unchecked; the gate keeps one witness per reason, checks that both entry points make the same sequence of compiler calls and throw the same errors, and compares the sending decision with the previous release on two independent rule pools. From 0.92.0 the allow list's whole tool names are checked under the tightening reading, because the engine's rule compiler no longer takes an allow reading at all: the guard proves the swap changes nothing on the wire — the published 0.90.0 build running on the engine generation that receives this list and the current build running on either engine generation emit byte-identical requests over two rule pools, every name the receiving generation refuses under the allow reading is refused by the current compiler under the deny reading and no other, and the mirrored rule shape's behavior set equals the installed compiler's. It also records why the engine's own allow-list grammar is not used as the judge yet: on the names this assembly checks, that grammar and the receiving generation's compiler disagree in exactly five registered classes (names with a backslash, which the receiving generation refuses and would reject the whole request for, are never sent), and a sixth class appearing turns the guard red. From 0.92.0 it also pins how path rules relative to the task root (`src/**`, `./x`, `/x`) are handled: they are sent exactly as written, byte-identical to every published build from 0.90.0 on (0.91.0 included, whose companion report is compared in full as well whenever the capability that turns allow rules on is not reported as true), and never rewritten to an absolute `//` form; for every server generation the gate is given, it records verbatim how that server's settings intake answers such deny and ask rules with and without a task root (refused without one, accepted with one, absolute `//` rules accepted either way), so a change on the server side turns the guard red. From 0.92.0 two of those classes (protocol names whose server segment carries a backslash or a wildcard) are held back before the name is checked at all, so three classes reach that judge, and the guard checks each held witness was never asked. Protocol rule names whose server segment the engine never mints as written (a wildcard segment, or one the engine's own naming function would rewrite) match no tool on any server generation: in the allow list they are held back with the reason `unminted_protocol_segment`, while deny and ask rules are still sent exactly as the previous release sent them (a tightening rule must not quietly disappear; from 7.116.0 the server refuses the whole request and names the rule) and are listed in the report's `unreachable` table with the prefix the engine mints for that server, so a client can warn before sending; the guard compares the deny and ask lists with the previous release byte for byte on a pool full of such names, and checks that the table lists exactly the sent rules the server's judge would refuse. Without the naming function installed the allow copies are held as not judged and only wildcard segments are listed. Since 0.94.0 the allow bucket follows the engine generation that receives it: with the allow-layer capability (and the engine's allow-list reader installed) content-form allow rules are sent as written, and an entry the engine would read as voiding the whole allow list is held back and reported; older engines keep the previous projection byte for byte. The one-sentence notice for the print lane when settings-file allow rules were held back because the settings read was not clean (`permissionsAllowKeptNotice`) is pinned here: the wording per read-health state, the managed-policy variant, no rule text in the sentence, malformed input treated as absent. 0.98.0: the deny / ask buckets now carry every non-empty rule (only equivalent rewrites; the report has no unchecked / unreachable tables; deny / ask kept rows are only empty_rule / duplicate), the allow bucket is byte-identical to the previous build, every throw from the assembly carries a machine-readable refusal identity (read from own data slots, identical in a minified bundle), and the live server fixture refuses / accepts each passed-through rule as expected. Since 0.99.0 the cells that use a bare wildcard or a padded name to reach the whole-list and re-check arms judge with the engine generation at the support floor, taken from the server fixture that bundles it (a missing fixture is reported as skipped, and is red in required mode); one more cell records how the newer installed engine refuses those two inputs before the run, with its own code and sentence passed through unchanged. |
 | `scripts/run-file-history-leg-notice-test.mjs` | What a client says after the engine reports, for one run, whether it is capturing file history. A five-way verdict combines the engine's capture capability as it was read when the request was assembled (not as it reads when the report arrives), whether this run's submitted request actually carried the per-run opt-out, whether the user turned file checkpoints off, and the word reported for the run; only one outcome produces a sentence (the run carried the opt-out and the engine still captures), with its conclusion first so a truncated status line still reads. The guard pins every one of the 54 cells of the verdict table, sweeps alternative spellings and malformed values for each input, checks that an unreadable capability never produces a sentence while a carried opt-out is judged without consulting the capability, and that inherited input properties are ignored. A process-local ledger records, per submission, whether the outbound request body carried the opt-out (only a plain own data property counts; a body with its own serialisation hook or an accessor for that key is read as unknown, so a fact is never recorded that the sent JSON could contradict) and binds the submission to its run only on a frame proving the run was created, using the same rule as the prompt-delivery ledger, so session-busy refusals and replayed approval cards carrying another run's id never attach this submission's fact to that run; answers do not depend on call order, conflicting records read as unknown, and runs this process never bound read as unknown. An end-to-end section sends real requests through the SDK stream parser to a local stub and drives the reported word through the projection and adapter layers to the verdict. A body whose property descriptor and property read disagree (a proxy), or that carries any enumerable getter, reads as unknown, because serialization runs those getters and may drop the field; a value that is a non-null object (a wrapped string, an object with its own serialisation hook) also reads as unknown, since what it serialises to cannot be proved. |
 | `scripts/run-support-floor-projection-test.mjs` | The package's machine-readable support floor: the oldest server, engine and SDK versions it still supports, kept in `scripts/support-floor.json` with the source of each value and exported at runtime as `SUPPORT_FLOOR` (with `supportFloorOf` and the engine version reading `engineVersionTier`). Since 0.95.0 the constant is not a hand-kept copy of the declaration but its build-time product: a generator writes one comment-free, import-free file from the declaration, and the check fails when the checked-in file differs from a fresh generation, carries a comment, when the runtime constant is not that very object, or when the source holds a literal version string; it proves itself against an unfaithful generator (a dropped key, a changed key order, a stale file, an injected comment, an orphan or missing file) and against twelve malformed declarations. The check also fails when the declaration cannot be read or is malformed, when the exported values differ from it, when the SDK value differs from the SDK peer range, and (when the full fixture set is required) when no server fixture in the list matches the declared server version or that server pins and installs a different engine than declared; every fixture in the list is printed with whether it sits below, at or above the floor. The version reading answers engine at and above the floor, legacy below it and unknown for anything that is not a plain release version (pre-releases, build metadata, a leading v, whitespace, wrong segment counts, leading zeros, the word unknown, non-strings); a caller-supplied threshold can only raise the floor, a malformed one is refused before any version is read, and hostile inputs never throw. The same reading exists for the version of the engine install's core package (`coreVersionTier`): the same parser, the same three answers and the same result type, differing only in where the lower bound comes from (the declared engine floor rather than the declared server floor); the check compares the two readings on one pool of inputs, pins the golden cases, and fails when either reading carries its own version comparison instead of calling the one shared function. Since 0.93.0 the check also pins that exactly one function in the whole source tree reads a version string (a planted version-shaped regular expression or split-and-parse is detected by the same scan); the per-feature version checks that used to carry their own thresholds — the limits and detach checks, the goal stop-hook tier, the plan-review three-choice tier and the inherited-environment server tier — no longer exist. Since 0.96.0 the check also covers the effective threshold read: the one value the engine version reading actually applies — the declared server floor or the caller's own floor, whichever is higher — is exposed as a single read, so a message that names the threshold prints the very version the check used. It pins that the read gives the same answer as the check on a pool of floors (the answer is exactly the lowest version the check lets through, including across segment boundaries), that a malformed caller floor is refused identically by both, that both come from one shared function whose body is a single delegation, that a version segment above 2^53 − 1 is treated as unreadable (a floor with one is refused, a version string with one reads as unknown) so the printed threshold and the admitted minimum cannot disagree, and that the generated declaration object is not part of the public surface. Since 0.98.0 the check also pins that the two outward copies of the floor — the README support-floor line and the support-floor row of the integration guide's version table — carry exactly the declared server, engine and SDK versions, so raising the floor in one place and not the others fails. |
 | `scripts/run-server-7113-fixture-test.mjs` | Adoption checks against the published server package, run once per server generation (the one that carries the change and the one just before it). Session reads, live steer and policy settlement are driven through the package's real routes and modules; the attachment checks read the package's shipped source text instead. Attachments: the keys the server's shipped shape description lists are exactly the keys this package knows, the server's shipped refusal for the retired plan-mode reminder key names its retirement, every value this package's attachment switch can produce stays inside the accepted set, and naming the retired key is refused locally instead of being sent. Session reads: a non-owner reading a session's detail, head or event stream gets a not-found whose code names the session (the previous generation named a run), and this package's session read answers the same way on both. Live steer: without an idempotency key the engine's input id equals the message id on the receipt, a queued frame and an applied frame on one input id each project as their own event with every field as written, and the delivery ledger records delivery once. Policy settlement: an unattended answer by class is recorded under the policy party as an allow settled by policy, read back verbatim, no longer read as a person's approval, and a tool-end frame carrying it still projects. |
@@ -496,7 +496,7 @@ guard still cross-checks the table by name).
 | `scripts/run-approval-respond-parking-test.mjs` | The live-card respond answer when the engine replies `202` with the reason `parking_in_progress`: the card's row is moving to parked, so the answer did not take effect. The outcome is `decision: 'unresolved'` plus a `parking` position carrying the approval id and, when the engine supplies them, the suggested retry interval and the decide path of the parked card; the resolution reader reports it on the existing unsettled arm with the new cause `parking_in_progress` (the union keeps its three arms; a retracted or edit-refused card still wins over it, and it wins over the decision word). The parking fact is carried only by an own data property naming the reason and a non-empty approval id that matches the card; a retry interval or decide path that is unusable drops only itself and the parking still stands. A body that names the parking reason but whose approval id is missing, unusable or belongs to another card is reported as not settled, never as decided and never as parked, with no notice and no resend. The answer is never re-sent, never has its attribution key dropped or renamed afterwards, and the package does not wait for the interval; no accompanying notice (for example the one saying the approval itself went through) and no attribution position is produced; exactly one user-facing sentence is shown (with the engine's suggested wait appended in seconds when given) and it contains no engine text. A 200 acknowledgement keeps its reading unchanged, bodies that are neither acknowledgements nor parking bodies are not read as parking, the durable decision path and its single holder have no parking cause, and a syntax-tree scan pins one reader declaration, one call site and the places where the reason word is minted, while the 202 body shape is pinned against every server generation in the fixture list. From 0.96.0 the base sentence is a public constant (`APPROVAL_PARKING_WARN_TEXT`, exactly the notice shown when the engine gives no wait hint, so a host that words its own parking notice no longer keeps a verbatim copy); the longer form with the wait hint and the notice function itself stay internal |
 | `scripts/run-decision-refusal-standing-test.mjs` | Shared decision-refusal reading for every client: `decisionRefusalAskReading` is the one reading of whether the question a refused decision was aimed at is already settled (settled, not settled, or unreadable), and settled comes only from positive evidence (a local no-pending stop, a 409 `conflict.ask_decided`, a `gone.ask` or a 410 other than a parked ask, a 404 whose body names an ending cause, or the matching error class when the code was stripped); a 404 without an ending cause, a not-found error class, a parked ask, the codes that only say the approval you looked at is no longer current, and wording in the failure text are not evidence. When the engine refuses an approval decision, `classifyDecisionRefusal` reports settled elsewhere exactly when that reading is settled, otherwise ambiguous (5xx except 501) or refused while the question may still be pending (every other 4xx and 501), returns nothing when no server verdict can be read, and marks transient refusals (408, 425, 429, 502, 503, 504, a body that says the request is retriable, or a 404 whose body says the id belongs to another replica); `liveAskPendingReading` reads whether the question is still in the engine's live pending list as pending, not pending or unknown (an unreported, malformed or partly dropped list is unknown); `decisionRefusalClosesOut` closes out only a refusal that is neither settled elsewhere nor transient and whose question the read confirms is still pending; `undeliverableDecisionContent` is the one sentence for each way the close-out ended. The guard covers the golden status/code table and near-miss codes, values that are not a verdict, guarded reads (throwing getters, prototype accessors, revoked proxies; the reading of the question itself takes every machine-readable slot of the failure from own data properties only: a value found only on the prototype chain is absent, and an own accessor makes the reading unreadable and is never run; the outcome note and the flattened outcome that also feed the reading take their status and error code from own data properties too, so a value on the prototype chain cannot be laundered into evidence), the outcome note shape, codes unsafe for display, the full close-out truth table and the exact sentences, and that the code literal and the classifier exist in exactly one place; it also checks that the classifier, the interrupt-deny check and the park-resume check give the same settled answer across 6912 failure shapes (the park-resume check also re-reads the stream when the approval it looked at is no longer current), the two shapes that used to get different answers, and that all three read the one shared reading. |
 | `scripts/run-no-allow-cause-test.mjs` | Why an automatic lane did not allow a call, and the refusal line for the two automatic-refusal lanes. `noAllowCauseOf(card, ladder)` combines the card's standing-rule verdict with the host's rule-ladder evidence (only an own, strictly `true` property counts): allowed or denied by a rule gives no cause; a local ask rule gives `ask_rule`; an exhausted automatic-approval budget gives `budget`; an ask rule the card names (its class or origin) gives `engine_ask_rule` when the ladder evidence shows that none of the user's own ask rules matched (the end-of-ladder no-match flag, or the dedicated flag for the case where an allow rule matched but the card's question class overruled it), and stays `ask_rule` when that evidence is absent, because the user's own rule cannot be ruled out; complete no-match evidence on a card a rule could clear gives `no_allow_rule`; everything else gives `other`, so no lane suggests adding an allow rule when that would not help. `policyDenyReason` mints the single line for the lane without an approval UI and for the dontAsk mode, one form per cause, naming only this one call and offering host-specific remedies (an allow-rule flag, a budget setting) only when the host supplies them. The guard covers the golden cases, pairwise ordering, strict reads, all ten sentences verbatim (the engine-side ask rule line says the user's settings cannot approve it and offers neither removing a rule nor adding an allow rule), unsafe tool names kept out of the sentence, and the single mint site. |
-| `scripts/run-approval-card-face-projection-test.mjs` | Three card positions added in 0.94.0. `boundGrants` (what approving a Skill card also grants: the skill name or null, the entries that then run without asking, and the entries refused for the rest of the run) is read once from the live approval frame through a strict reader; any malformed part drops the whole position, the copy is frozen, the digest is not put on the card, and the durable pending-row path never carries it; `approvalBoundGrantsDetail` gives one sentence with counts only, never the deployment text. `classifierNoRuling` (why the auto-mode classifier gave no ruling) is read by word shape as an open set from the live frame only; `classifierNoRulingDetail` has a phrase for each of the engine's words and a generic sentence otherwise. `priorCardsForCall` lists, for the same tool call, the earlier cards that asked about something else (a different class set, origin or mandate word), each with its approval id, origin, mandate word and, when answered here, whether it was allowed or denied; replays and parked twins of the same question do not count, both card paths share one bounded ledger, and `approvalPriorCardsDetail` says this is a separate check that needs its own answer. |
+| `scripts/run-approval-card-face-projection-test.mjs` | Three card positions added in 0.94.0. `boundGrants` (what approving a Skill card also grants: the skill name or null, the entries that then run without asking, and the entries refused for the rest of the run) is read once from the live approval frame through a strict reader; any malformed part drops the whole position, the copy is frozen, the digest is not put on the card, and the durable pending-row path never carries it; `approvalBoundGrantsDetail` gives one sentence with counts only, never the deployment text. `classifierNoRuling` (why the auto-mode classifier gave no ruling) is read by word shape as an open set from the live frame only; `classifierNoRulingDetail` has a phrase for each of the engine's words and a generic sentence otherwise. `priorCardsForCall` lists, for the same tool call, the earlier cards that asked about something else (a different class set, origin or mandate word), each with its approval id, origin, mandate word and, when answered here, whether it was allowed or denied; replays and parked twins of the same question do not count, both card paths share one bounded ledger, and `approvalPriorCardsDetail` says this is a separate check that needs its own answer. Each line of the card's "why is this asking" block is rendered from the package's own sentence; only the values interpolated into a sentence (a limit name, a rule entry, probe text, an engine message, a cause word or code) go through the display sanitizer, so a fixed sentence such as "only its basic facts are shown" is no longer rewritten by the credential filter, while a credential-shaped value in any interpolated slot is still masked and never shown in the clear. |
 | `scripts/run-hitl-words-readout-test.mjs` | The five wording and reader fixes of the HITL surface. (1) Seven user-facing failure sentences (pending-row fetch, approve and deny decisions on approval cards, the question-card decision, the parked-row wait, local rule parsing, background stop) carry the thrown value's own message and never the error class name — a thrown value with no usable message becomes one fixed sentence instead of a bare class name, `[object Object]` or `null`; the original thrown value goes only to the host log. (2) `resumeContextUnavailableContent` is a single English source (three ` · ` segments, with and without a run id, the retention seconds only when the engine gave them) and a plan-review decision that meets a 409 `conflict.resume_context_unavailable` says it. (3) The plan-review `permissionModeAfter` pre-check no longer mirrors the server's decision pairing: only the value set is checked locally, and the engine's `request.field_conflict` speaks for the rest. (4) `requestEngineCompact` reports how the fire settled through `onSettled` (accepted / refused / failed / offline), with a 409 `compact.not_running` also surfaced on the host surface. (5) Every exported `project*Capability` reads only own data descriptors: an accessor is never run and reads as `undefined`, and a throwing descriptor read is swallowed the same way. Pinned with throwing and class-named values at all seven sites, both plan-review entry points, the compact outcomes under 409 / 404-retry / 5xx / the 60 s injected deadline, and a per-projector throwing-proxy row plus a source scan. The decision exits treat hostile thrown values uniformly: a throwing message getter, a throwing toString, a revoked proxy, or throwing status, error-code or name getters all end in the thrown value's own message or the fixed fallback sentence, never in the getter's own error text, and the transient-failure classification reads each slot independently. |
 | `scripts/run-output-style-seat-test.mjs` | The per-turn output-style seat. From 7.121.0 the request accepts a top-level `outputStyle` object (`name` and an optional `turnReminder`) that the engine states once per attachment cycle, and the capabilities body carries a constant top-level bit for it; the reader keeps the same three states as its constant-bit siblings and does not confuse it with the nested bit for the style body. The live-defaults layer places the host's `outputStyleSeat` on the request only when that reading is admitted, copies it as given (an empty name or extra keys are passed through for the engine to refuse at prepare time with `config.output_style_invalid`, which the configuration-refusal table recognises), never overwrites a caller's own top-level value, never invents one, and leaves the style body seat unchanged; the field is registered in the live-defaults list, survives the unregistered-key check on both lanes, and is reported as belonging to another channel when passed to the request assembler. |
 | `scripts/run-request-uplift-test.mjs` | Six request-assembly and reader pieces shared by every client. The per-run declaration that nobody on an unattended run can answer approval requests is carried only on the unattended lane, only when the engine reports it accepts that field (an older engine rejects the whole request), with one plain sentence for each reason it was held back; the headless permission mode is always stated explicitly, falling back to `default` when neither flag nor environment sets it; a predicate tells apart the denial an engine makes when nobody is reachable and no queue can hold the request; a host-reported route problem (no gateway address, or the anthropic channel with no credentials) turns a model that has a name but cannot start into a stated configuration problem; a reader gives, per model, which thinking levels the engine accepts; attachments from MCP resources and the new not-attached reasons are written as data envelopes rather than as tool calls; and servers the client mounts itself state their own allow rule, audited separately from the user's rules, with the rule-source trust order exported once. |
@@ -505,8 +505,8 @@ guard still cross-checks the table by name).
 | `scripts/run-seat-wave2-test.mjs` | The seat capability table, the folder picker and the session folder, added in 0.94.0. Whether a renderer may offer an optional seat action is answered only by the capability table the host publishes (`seatCapability`: reported, not reported, or unreadable — reading own data properties only, never running a getter), never by calling the action and watching it fail; "not reported" is neither folded into "available" nor into "cannot", and the disabled sentence comes from one place (`seatCapabilityUnavailableDetail`). The subscription helper subscribes before it reads, so a renderer that subscribes after the host has published and then withdrawn a capability sees the withdrawn state on its first callback. `pickFolder` returns a non-empty path or `null` for a cancel, and the session record's `cwd` is the folder the session was started in, absent when none was chosen. The same gate pins `deviceFlowPollingPlan`: the device-code response's own positive finite numbers win, anything missing or malformed falls back to the contract numbers generated from the settings schema at build time (no number is written by hand), and the poll interval is never below one second |
 | `scripts/run-wiring-instructions-attribution-test.mjs` | Which project instruction file the engine actually loaded for a run, attributed to the client's own directory. The capabilities bit `callerCwd` gets its own reader (the same four ports as every capability reader; `callerCwdAdmitted` accepts only an explicit `true`), and a host that notes the bit inside its probe callback reads the new value in the same tick. The working directory a submitted request carried is recorded per run in the same in-process ledger and with the same frame-binding rule as the file-history opt-out (`legCwdOfBody`, `declaredLegCwd`, `legCwdOf`). `engineSelectedInstructionsFile` then returns the selected member only when the run's directory equals the client's current directory, the member is a plain relative path (no `..`, no absolute or drive form, no backslash, no control, bidi or invisible characters), and the host's real-path function confirms it stays inside that directory (a symlink pointing outside is refused). |
 | `scripts/run-layering-direction-test.mjs` | The package's layering table (`docs/LAYERING.md`) and its direction gate. Every source file sits on exactly one internal layer (wire readers and injected ports, projection, host-facing orchestration, the two entry points) and may import only downward, type-only imports included; the exemption table only shrinks. Every import that leaves the package must match a row of the cross-repository table and the kind that row allows: the engine package is read for types only and never at runtime, and the shipped declaration files reference only the packages listed in `dependencies` and `peerDependencies`. |
-| `scripts/run-panel-reduce-test.mjs` | The panel row reducer: one record per background agent, found through one identity index of engine ids — the fleet row id tail and the sub-run id (the id ticks and transcript anchors carry) are bound together only when a single frame states both, and two rows later proven to be the same agent are merged, the duplicate leaving with a `remove` that names the surviving row (`supersededBy`). A delegating card id never decides which agent a frame belongs to: it is a display field and an "already seen in this process" hint only. One generation rule, one place that constructs terminal facts and one place that emits a `terminal` row delta. Usage is monotonic within one known generation: a tick (which carries no generation) moves the numbers only for an agent without cycle numbers, a row frame that binds a new sub-run id to the same tail starts a new generation, a terminal row freezes the phase but not the usage, and finality is reported per counter (`usage_delta.final`), a value-unchanged change of finality included. `reducePanelFrame(state, frame)` is pure — no I/O, no clock read (the projection frame carries its own `nowMs`), no module state, the input state is never mutated, the same input always yields the same deltas and replaying a frame yields none. The delta set is closed (`upsert` / `usage_delta` / `terminal` / `remove`); `subscribeEnginePanelRows` replays a snapshot on subscribe. Twelve end-to-end sequences cover interleaved lanes, the first-published id, revival, reconnect/absence, late subscription, short-lived agents in both orders, historic rows, absent usage never filled with zero, the turn-end sweep scope, the transitional event channel versus the row channel, bounded eviction and snapshot fidelity. The generation criterion now lives in a dependency-free leaf module that both the panel record and the fleet ledger share; each layer keeps its own state (the panel's per-row watermark, the ledger's per-task watermark) but the comparison and the notion of a valid cycle number exist once. A data-flow scan keeps cycle numbers and watermark readings out of every comparison operand in the ledger (with planted violations to prove the scanner speaks, and no false alarm on presence checks or shape tests), and the numeric view is pinned against an independently written integer comparison over a grid that includes malformed values. |
-| `scripts/run-sdk-witness-compile-test.mjs` | The package source compiled against every SDK generation the peer range promises: each installed SDK at or above the declared peer floor (read from `package.json`, not hand-pinned in the guard) is pointed at through the compiler's `paths`, with the package's own `tsconfig.json` options, and `src` must produce zero diagnostics against each one. The ruler is proven to speak on every generation: the program covers every `src` file found by an independent directory walk, every SDK import in `src` resolves into the generation under test rather than the development dependency, and an in-memory planted call that omits the required `baseVersion` and a status word outside the closed union must both be reported while the corrected call stays silent. A probe tree installed on disk but missing from the witness list is red; an absent probe set is an honest skipped section, never green. |
+| `scripts/run-panel-reduce-test.mjs` | The panel row reducer: one record per background agent, found through one identity index of engine ids — the fleet row id tail and the sub-run id (the id ticks and transcript anchors carry) are bound together only when a single frame states both, and two rows later proven to be the same agent are merged, the duplicate leaving with a `remove` that names the surviving row (`supersededBy`). A delegating card id never decides which agent a frame belongs to: it is a display field and an "already seen in this process" hint only. One generation rule, one place that constructs terminal facts and one place that emits a `terminal` row delta. Usage is monotonic within one known generation: a tick moves the numbers when its own cycle number matches the row (for an agent without cycle numbers the sub-run id decides), a row frame that binds a new sub-run id to the same tail starts a new generation, a terminal row freezes the phase but not the usage, and finality is reported per counter (`usage_delta.final`), a value-unchanged change of finality included. `reducePanelFrame(state, frame)` is pure — no I/O, no clock read (the projection frame carries its own `nowMs`), no module state, the input state is never mutated, the same input always yields the same deltas and replaying a frame yields none. The delta set is closed (`upsert` / `usage_delta` / `terminal` / `remove`); `subscribeEnginePanelRows` replays a snapshot on subscribe. Twelve end-to-end sequences cover interleaved lanes, the first-published id, revival, reconnect/absence, late subscription, short-lived agents in both orders, historic rows, absent usage never filled with zero, the turn-end sweep scope, the transitional event channel versus the row channel, bounded eviction and snapshot fidelity. The generation criterion now lives in a dependency-free leaf module that both the panel record and the fleet ledger share; each layer keeps its own state (the panel's per-row watermark, the ledger's per-task watermark) but the comparison and the notion of a valid cycle number exist once. A data-flow scan keeps cycle numbers and watermark readings out of every comparison operand in the ledger (with planted violations to prove the scanner speaks, and no false alarm on presence checks or shape tests), and the numeric view is pinned against an independently written integer comparison over a grid that includes malformed values. A background agent terminal notification may carry the child session id; the reducer treats it as binding evidence, so a host that feeds only task events (no fleet rows) gets one row per child with the engine's final numbers, and a revived child no longer keeps the previous generation's tool count beside the new generation's token count. A tick carries its engine cycle number: a background child usage is driven by ticks again when the tick cycle matches the row, a tick from a previous cycle is dropped whole, a tick from a newer cycle neither raises the watermark nor counts, and a terminal tick marks exactly the counters it really carried as final. |
+| `scripts/run-sdk-witness-compile-test.mjs` | The package source compiled against every SDK generation the peer range promises: each installed SDK at or above the declared peer floor (read from `package.json`, not hand-pinned in the guard) is pointed at through the compiler's `paths`, with the package's own `tsconfig.json` options, and `src` must produce zero diagnostics against each one. The ruler is proven to speak on every generation: the program covers every `src` file found by an independent directory walk, every SDK import in `src` resolves into the generation under test rather than the development dependency, and an in-memory planted call that omits the required `baseVersion` and a status word outside the closed union must both be reported while the corrected call stays silent. A probe tree installed on disk but missing from the witness list is red; an absent probe set is an honest skipped section, never green. Type pins that need an SDK newer than the floor live outside `src` in a pin file named after the lowest SDK version they need; every witness at or above that version compiles the pin file in the same program, each named pin group must be free of diagnostics, and an in-memory false pin written with that generation's real types must be reported while its true control stays silent. The published declaration files, emitted in memory from the development SDK, must type-check with zero diagnostics against the witness at exactly the floor version (absent: an honest skipped section), so the public surface never names a type the floor generation lacks; a planted declaration that names a type only the newer SDK exports must be reported against the floor and stay silent against the development SDK. |
 | `scripts/run-interactive-interrupt-emitter-test.mjs` | The interactive interrupt sender (`fireInteractiveInterrupt`): one bare `POST /v1/runs/:id/interrupt` through the client's runs resource, with the method called on its receiver, the session parameter sent only when non-empty and no idempotency key, so exactly one request goes out. Every outcome is a value, never a throw: a receipt yields `halted` with `turnCut` read only from an own boolean data property; a thrown error is classified by the same at-most-once judge the steer and cancel paths use, so a definite refusal becomes `refused` with its status and code, while a 5xx, a network failure or an abort becomes `transport` (no verdict). A missing resource yields `unarmed` with zero requests. The suite cross-checks every thrown sample against the shared judge, drives a real SDK client with a stubbed fetch and feeds the outcomes to `planInteractiveHalt`. |
 | `scripts/run-device-rebind-judgement-test.mjs` | The device rebind flow (`listDevices`, `readSessionBinding`, `rebindSession`, `reconcileRebindOutcome`, `rebindProvesRequestNotApplied`): each verb goes through the client's devices resource and returns a discriminated outcome, never a throw. Refusal codes are matched as status-and-code pairs from one closed table (device lane absent, run store absent, not found, active run, revision conflict, service token required); any other failure of the rebind write is classified by the shared at-most-once judge into `refused` or `unconfirmed`. A rebind success is accepted only when the echoed session, the target device and the revision step all hold, and the workspace carryover disclosure is carried only for the literal `none`. Response fields are read as own data properties. Includes a real SDK client with a stubbed fetch and a full grid of the read-back reconciliation. The eleven device row types from the SDK (`DeviceSummary`, `DeviceBinding`, and the rest) are re-exported as types from the package root, and the guard checks each one is identical to the SDK's. |
 | `scripts/run-engine-env-sourcing-test.mjs` | The local-engine environment group sourcing judge (`resolveEngineEnvSourcing`, `engineEnvView`): when a settings source wrote half of an all-or-nothing group and that half is ignored, the group is either skipped (its own switches are dropped) or, for safety groups and groups declared not skippable, left for the engine to refuse with an explanation. Group declarations, the ignore rule and the alias table come from the host; the module holds no retired-name table and reads no platform state. The suite runs a seeded differential test against a frozen transcription of the previous host-side loop, fixed cases for alias precedence, source precedence and case folding, and scans the built module for every retired environment name in the settings contract. |
@@ -533,7 +533,7 @@ guard still cross-checks the table by name).
 | `scripts/run-prompt-blocked-sentence-projection-test.mjs` | The sentence for a run whose first prompt was held back by a deployment's `userPromptSubmit` hook. The engine frames its own explanation for the model in an envelope that carries the session's reminder mark, and the terminal row used to print that envelope verbatim (`Run blocked: <system-reminder mark="…">…`) while the machine-readable cause beside it went unread. The package now has one sentence-maker for the `blocked` terminal, shared by the terminal row and the result frame's first error: a hook that refused the prompt reads `UserPromptSubmit operation blocked by hook:` followed by the hook's own words (treated as untrusted text — control and bidirectional characters shown as visible escapes, credentials washed, folded to one line, envelope tags removed even when spelled with zero-width characters); a hook that crashed, timed out, or was cancelled mid-check gets one fixed sentence each, saying the prompt was not submitted; an unrecognised cause word prints only the lead and never echoes the word. When the machine-readable cause is missing or malformed, the old `Run blocked: <reason>` line stays, with a whole engine envelope unwrapped so no tag or mark reaches the screen, and an agent's own free-text reason is byte-for-byte what it was. The new line counts as a governance stop (it is not an `API Error:`) and is recognised by the same row predicate as `Run stopped` / `Run blocked`. The suite also pins the single mint site in source, covers the interactive lane, and checks the tag remover is linear on near-miss input. |
 | `scripts/run-fleet-terminal-retain-test.mjs` | The fleet ledger's rule for terminal rows has one home and one function. A terminal row (completed, failed, killed) lives only in the retention pool and never in the active set, whichever way it arrives — a task frame or a reconnect snapshot — and both arms go through the same admission function. A row brought back by a reconnect snapshot therefore expires with the retention window exactly like one that arrived as a frame; before, the snapshot arm left it in the active set for good, and a lost removal frame did the same to a frame-delivered row, and clearing the pool could not clear it. The suite refutes the change on three axes: timing (snapshot before or after the terminal frame, interleaved with removal, repeated delivery, resurrection, a later snapshot that does not carry the row, a row being viewed, pool clearing, and every terminal word through both paths with identical answers), concurrency (twin rows of one task in either order, a notification-retired entry meeting a snapshot row, notifications against a snapshot-carried row, two sessions' ledgers), and failure class (a status outside the terminal vocabulary, absent, or null is never turned into a terminal row; rows without a usable id, with a non-string id, or null entries are dropped and counted, never thrown on). It also pins that the removal hook still reports a known row for a settled terminal row, because downstream bookkeeping credits a removal only when neither the stale mark nor the unknown-row mark is present. Finally it pins the terminal evidence the ledger keeps past the window (status word plus cycle, per task tail, bounded): after a row has aged out or the pool was cleared, a same-cycle completed notification cannot turn a known failure or kill into a success, while a newer cycle, a revival, and the upgrade direction still pass; the evidence itself is cycle-monotonic, so an older generation's terminal row, or one arriving late after a revival, neither overwrites nor re-plants it. Since 0.97.0 it also pins the ledger's single per-task rule, applied in the one admission function to task frames and snapshot rows alike: a reading from an older cycle than the task has already shown is ignored (it neither revives the row, nor withdraws the terminal evidence, nor lets a snapshot drop the task's newer in-flight row); a newer cycle retires older in-flight aliases of the same task; within one cycle, or when the order cannot be told, an in-flight alias wins over a terminal row; otherwise the terminal row settles the task and twin rows collapse into one. Every arrival order of a set of readings, through either arm, gives the same projection and the same in-flight reading. A notification from an older cycle than the task has shown leaves the rows alone even when the running row's latest frame carried no cycle, and the bounded per-task cycle record never evicts a task that still has rows. |
 | `scripts/run-bg-subagent-row-test.mjs` | The background-subagent identity predicate every end shares. The fleet read surface lists every non-terminal task row — and since server 7.125.0 a session-scoped stream also brings back the session's own run row and its foreground delegation rows — so "is this row a background subagent" needs exactly one answer. `isBackgroundSubagentRow` applies four rules in order: a row with no id, a composite id, or a `run-leg` lane mark is never a subagent (the last two are foreground delegation rows); a non-agent handle (background shell or monitor) never is; a single-segment row with a readable parent is; a row without a readable parent is only when its id is an agent-family handle. `isBackgroundWorkRow` adds the non-agent handles for the "N background tasks" counts, and `isOwnBackgroundSubagentRow` adds ownership — a session-scoped stream replaces the ownership half only, never the row-kind half. The handle shapes are not retyped here: the durable agent-handle pattern is the installed engine's own, and the non-agent prefix letters are read from the engine's task registry when the tables are generated. The read-surface row now carries `sourceLane` (only when the wire row has a non-empty string), and task frames and reconnect snapshots reach the predicate through the same row shape. The suite checks each rule with a positive and a negative example, the absence cases (no id, unreadable parent, unusable lane mark), the 7.125.0 stream shape against the older shape, and that the lane word and the prefix table each have exactly one home in the source. |
-| `scripts/run-display-credential-reach-test.mjs` | The check for how far the display credential masking reaches. Percent-encoded text is read through decoded views inside the same single scan (up to three layers, on the raw text and on the painted screen text), so an encoded separator such as `token%3D<key>`, an encoded label pair, an encoded separator followed by whitespace before the value, or a whole URL that was percent-encoded once or twice is masked exactly like its plain form, with the mask placed on the original escaped bytes and the surrounding diagnostic words kept byte for byte. A known key prefix right after a slash (`GET /sk-…`, a multi-segment request path, or its encoded form) is now masked too, while segments inside a `file:` address that starts its own token (a `file:` buried inside a request path or query value does not count), short directory names that merely start with a prefix, and random-looking path segments are left alone. The configured-address display and the upload check read one value judge: a query value is masked or reported when its key is a credential label, its value is credential-shaped or carries a nested credential (including encoded layers and `+` as a space), and readable word chains under non-secret keys such as `deployment` stay visible on screen but still count as credentials for upload. A sixty-eight address pool (including single-slash spellings, which are rebuilt from the parsed components first, also when a query value holds a nested address), a seventeen query pool, the rule that anything the configured-address display would mask is also reported as a credential for upload, idempotence on dot, space and unfolded exits and the absent-input rules are pinned, and when a terminal reference tree is present both functions are compared with the terminal's own implementation byte for byte. Since 0.98.0 the configured-address display and the upload check read the whole string as one address: a credential value keeps its tail after quote, bracket, pipe or backslash characters inside the mask, a token split by a tab or line break is judged as the URL parser sees it, a relative reference starting with a slash, question mark or hash has its query values masked, and a host name that looks like a credential label next to a numeric port is shown as written. |
+| `scripts/run-display-credential-reach-test.mjs` | The check for how far the display credential masking reaches. Percent-encoded text is read through decoded views inside the same single scan (up to three layers, on the raw text and on the painted screen text), so an encoded separator such as `token%3D<key>`, an encoded label pair, an encoded separator followed by whitespace before the value, or a whole URL that was percent-encoded once or twice is masked exactly like its plain form, with the mask placed on the original escaped bytes and the surrounding diagnostic words kept byte for byte. A known key prefix right after a slash (`GET /sk-…`, a multi-segment request path, or its encoded form) is now masked too, while segments inside a `file:` address that starts its own token (a `file:` buried inside a request path or query value does not count), short directory names that merely start with a prefix, and random-looking path segments are left alone. The configured-address display and the upload check read one value judge: a query value is masked or reported when its key is a credential label, its value is credential-shaped or carries a nested credential (including encoded layers and `+` as a space), and readable word chains under non-secret keys such as `deployment` stay visible on screen but still count as credentials for upload. A sixty-eight address pool (including single-slash spellings, which are rebuilt from the parsed components first, also when a query value holds a nested address), a seventeen query pool, the rule that anything the configured-address display would mask is also reported as a credential for upload, idempotence on dot, space and unfolded exits and the absent-input rules are pinned, and when a terminal reference tree is present both functions are compared with the terminal's own implementation byte for byte. Since 0.98.0 the configured-address display and the upload check read the whole string as one address: a credential value keeps its tail after quote, bracket, pipe or backslash characters inside the mask, a token split by a tab or line break is judged as the URL parser sees it, a relative reference starting with a slash, question mark or hash has its query values masked, and a host name that looks like a credential label next to a numeric port is shown as written. Since 0.100.0 the check also pins two display-exit fixes: a credential-looking word inside a host name (`my-api-key:443`) keeps its port and path in every combination of credential nets, including the words-only one, and a non-ASCII percent escape (a malformed byte sequence, or the next byte of a valid UTF-8 character) no longer hides a known key prefix right after it, in a path segment, a query value, an address inside free text, a bare token and the upload check; when a terminal reference tree is present, a 27-cell stop-character census, a 16-cell malformed-escape probe and a 216-cell pool are compared with the terminal exits and with the terminal settings check. |
 | `scripts/run-notice-presentation-test.mjs` | Presentation facts for engine notices, shared by every client. The cut primitive (`cutForDisplay`) shortens a sentence to a budget at a sentence boundary, then a clause, then a word, then a character, keeping at least half of the usable length at each step, counting the mark inside the budget, never splitting a surrogate pair or an escape token, and leaving the text untouched when the budget is not a finite number. The facts reader (`noticePresentationFacts`) answers who a notice is for (user and operator from the generated audience table; unknown for any code outside the catalogue), whether it carries something the person can do (present only when `detail.remedy` is an own non-empty string, otherwise unknown; absent is never invented), the remedy text, whether repeats are the same occurrence (only for the seven codes whose contract names an identity key, checked against the installed engine contract; every other case is an event), and a seat class (persistent with a hint, persistent, transient or diagnostic; `NOTICE_SEAT_CLASS`, `noticeSeatClassOf`). `outletReachable` and `planOutletRendering` decide which ways out a sentence may name in a given lane and deployment: reachable ones are written, unreachable ones are dropped, unreadable ones are not asserted. `noticePrintLine` builds the one-line print-mode form `[sema] <sentence> [<code>]` with a per-audience budget, sentence-boundary cut, a full-text pointer only when a session transcript exists, no cut when there is nowhere to point to, a single line, and the shared credential and control-character wash. The suite runs the real built package on each reader, probes hostile inputs and getters, checks the identity table against the installed engine contract, and reads the source to confirm there is one implementation of the surrogate and escape boundary. A cut never lands inside a redaction marker (the marker is removed whole), including on the 8000-character defence bound. |
 | `scripts/run-print-permission-outcome-test.mjs` | The one-line permission outcome for print mode (`printPermissionOutcomeOf`). It reads only the engine's own outcome for a tool call (the gate disposition and its settlement) and answers one of three forms from a single sentence table: the permission mode denied the call (the dontAsk sentence names the mode and offers an allow rule; the bypassPermissions sentence says it could not tell whether the write lands on a protected path; an unknown or missing mode word gets a generic sentence that never echoes the word), nobody could be asked (print mode has no approval channel; the sentence no longer claims a mode), or the mode let a call run without asking (the sentence for the recorded reason, shared with `modeReleaseReasonSentence`). Every other outcome, a missing record and hostile input answer nothing. The notice is declared for the user with or without a way out, and its seat class is read from the shared seat-class table; combined with `noticePrintLine` it forms the line `[sema] <sentence>`. |
 | `scripts/run-engine-directory-test.mjs` | The read faces for the cloud control plane's two engine endpoints: the directory of engines visible in a scope, and the "my engine" resolution a client asks for right after login. Four things are guarded because each one is a way to be confidently wrong. An unreadable response is not an empty directory: only an explicit empty list means "no engines here", and a body that is missing the list, or is not an object at all, voids the reading rather than reporting zero. A row that cannot be read is dropped without taking its neighbours with it, but the roster then says how many rows it could not read, so nobody treats it as complete. The relay path on a row is the one string the client appends to the center's origin and sends the user's token to, so it is accepted only as a same-origin relative path — an absolute URL, a protocol-relative form, one that only becomes protocol-relative once the URL parser has eaten a tab or newline, or one with a dot segment (plain or percent-encoded) that the parser would fold into a different path makes the whole row unreadable instead of being cleaned up and let through. And the words the server may add later are read as an open set: an unlisted kind, posture or resolution rule is passed through verbatim rather than thrown on or folded into a known word, while a missing online flag stays "not reported" instead of becoming "offline". The decision on top of the reading never guesses: a target the server named is used only when the candidate roster is present and agrees with it, a target the client could not read is never turned into "no engine", a resolution rule this version does not know with no target neither claims "no engine" nor "ambiguous", and a body whose rule contradicts its target or roster is not acted on. The explicit engine override is checked against the candidate roster only, by exact bytes; a name that also appears on a row that could not be read counts as a duplicate rather than disappearing with the filtered row; and when the roster itself cannot be read the override is refused rather than falling back to connecting by name |
@@ -542,7 +542,7 @@ guard still cross-checks the table by name).
 | `scripts/run-seat-settings-default-model-test.mjs` | The seat settings state no longer carries the deployment's default model. That fact has one reader, `liveDefaultModelOf`, applied to the model catalog the seat's `listModels()` returns; a second copy published through the settings store (a different wire field, taken at a different moment) could only disagree with it. The settings state now holds just the engine version and the data root, and a host that still publishes a `defaultModel` key is refused at the outbound gate, with the reason pointing at the catalog. The guard also reads the built declaration for the exact member list, and pushes six catalog bodies (default found by name, by id, absent from the catalog, key missing, key malformed, two rows sharing an id) through the body reader, the seat gate and both default-model readers to show they agree byte for byte. |
 | `scripts/run-seat-recents-test.mjs` | The seat's recent folders. The recent-folders list is read in exactly one way: the folders of the session records (derived, never registered by the host) joined with the folders the host registered (the ones the user picked, whether or not a session ever opened there), de-duplicated by exact path, most recent first, and a path present on both sides counts once at its later time. `recentFoldersOf` is the pure reader; `readRecentFolders` is the orchestrating reader that calls `listRecentFolders` only when the seat capability table says `recents` is available, and that keeps the session-record half when the host half is not offered or cannot be read. The seat gets two optional verbs, `listRecentFolders` and `noteRecentFolder`, and one capability bit `recents` that governs both (an absent bit means the host did not say, never "none"). The guard checks the registration, the outbound gates row by row (plain objects only, so what passes the gate is what survives structured cloning), the capability reading and its three distinct sentences, the reader's absence and ordering rules (a reply whose rows are all unreadable is reported as unreadable, never as an empty list), and an end-to-end chain from the host's capability publication through the renderer's reading to the host's verb double. |
 | `scripts/run-approval-card-v1-projection-test.mjs` | Projection of the v1 approval card (the `card` of an `approval_request` frame) into approval-card request facts through `approvalCardFactsOfV1`. For every key the v1 card shares with the live approval frame, the same value, well-formed or malformed, fed to both paths yields the same card request key and value; rule offers land in the read-only slot because the ask-decision route carries no rule; the real-approval bit is read through the single real-approval reader (top level or `risk`); every key in the SDK card key list has a recorded disposition, so a new SDK card key fails the gate; keys outside the card contract are not read, host positions are not minted and the session-answer route fact is not stamped; the grant-binding state matches the live path; refusal evidence that is present on the card but unreadable (malformed, inherited or throwing) is recorded as `unreadableEvidence` identically on both paths; non-objects yield undefined, one throwing key drops only that key and records it, and the result is frozen; the v1 card, live-frame and generic-card projections answer every shared key identically. |
-| `scripts/run-standing-rows-test.mjs` | Single judge `standingRowsOf(card, host?)` for the three standing rows on an approval card: allow for this session, don't ask again, and the deployment-list session row. Every row widens what runs without asking, so each one fails closed. An unreadable card, `rememberRefused` from the engine (own or inherited) or a present-but-unreadable value closes all three rows, even when rule offers are present. The session row needs the route fact `allowSessionCarried`, which the live-frame and pending-row paths stamp and the v1 path does not. The deployment-list row needs every ask class to be the deployment-list class, checked against the installed engine's `classifyAsk`. Engine rule offers in either slot give `engine_offer`; an unreadable offer key gives none. A local whole-tool rule is offered only when no engine lane closed the door, the arguments are available, the grant binding reads `none`, a standing allow rule could clear the ask, and the host reports a generic card face and that the rule would be sent; each precondition is flipped on its own. The live-frame, pending-row and v1 paths are exercised end to end, the grant-binding reading (since 0.98.0 the engine's own `boundGrantAbsence` word, not the tool name) is compared form by form on all three paths with the server's three grant-binding functions when a 7.129.0-or-later server fixture is installed, and may only be stricter where the card cannot tell which of them the server applies, malformed, inherited or throwing refusal evidence on the live-frame, pending-row and v1 carriers is recorded on the card as `unreadableEvidence` and closes the rows instead of reading as absent (checked end to end through the real paths); results are frozen, and hostile host facts never throw. Since 0.97.0 the seat request's allow-for-this-session bit `sessionRow` is minted by `toolPermissionRequestAskBits(card, requestId)` from the same premise as the judge's session row and the request id's domain (`live:` and `durable:` carry a session answer; `question:`, `elicit:` and `plan:` do not); an unreadable or refused remember flag never lets the seat show that row, and the crossing never copies the bit from its input — a seat request, or any input carrying a request id, passed back through it gets no bit. |
+| `scripts/run-standing-rows-test.mjs` | Single judge `standingRowsOf(card, host?)` for the three standing rows on an approval card: allow for this session, don't ask again, and the deployment-list session row. Every row widens what runs without asking, so each one fails closed. An unreadable card, `rememberRefused` from the engine (own or inherited) or a present-but-unreadable value closes all three rows, even when rule offers are present. The session row needs the route fact `allowSessionCarried`, which the live-frame and pending-row paths stamp and the v1 path does not. The deployment-list row needs every ask class to be the deployment-list class, checked against the installed engine's `classifyAsk`. Engine rule offers in either slot give `engine_offer`; an unreadable offer key gives none. A local whole-tool rule is offered only when no engine lane closed the door, the arguments are available, the grant binding reads `none`, a standing allow rule could clear the ask, and the host reports a generic card face and that the rule would be sent; each precondition is flipped on its own. The live-frame, pending-row and v1 paths are exercised end to end, the grant-binding reading (since 0.98.0 the engine's own `boundGrantAbsence` word, not the tool name) is compared form by form on all three paths with the server's three grant-binding functions when a 7.129.0-or-later server fixture is installed, and may only be stricter where the card cannot tell which of them the server applies, malformed, inherited or throwing refusal evidence on the live-frame, pending-row and v1 carriers is recorded on the card as `unreadableEvidence` and closes the rows instead of reading as absent (checked end to end through the real paths); results are frozen, and hostile host facts never throw. Since 0.97.0 the seat request's allow-for-this-session bit `sessionRow` is minted by `toolPermissionRequestAskBits(card, requestId)` from the same premise as the judge's session row and the request id's domain (`live:` and `durable:` carry a session answer; `question:`, `elicit:` and `plan:` do not); an unreadable or refused remember flag never lets the seat show that row, and the crossing never copies the bit from its input — a seat request, or any input carrying a request id, passed back through it gets no bit. The deployment-list question class names come only from the generated ask-class table. |
 | `scripts/run-refusal-machine-reads-test.mjs` | Machine-readable causes on refusal outcomes, shared by every client. A rule the engine's rule compiler refuses is reported in `permissionsDeclarationReport().kept[]` with the compiler's own `code` and `message` for that rule (present only when readable as non-empty text; other reasons never carry them), so a host does not ask the compiler a second time. The refusal note on a live approval outcome (`respondRefusal`) keeps every machine-readable part the shared refusal readers use: the server's `retriable: true`, the not-pending cause word and the refused field from the refusal body (`extra`), and an `unreadable` mark when the thrown error itself cannot be read, so `classifyDecisionRefusal`, `decisionRefusalAskReading` and the close-out check answer the same for the thrown error and for the note. A binding safety stop says where it came from in its code: `binding_missing` when the pending row lacks the binding fields and nothing was sent, `binding_mismatch` only when the engine answered with a binding conflict; the closed set of safety stop codes is frozen, and the shared readers treat the new code as a local stop (not settled, never read as already resolved). 0.98.0: compiler refusal codes are pinned on the allow bucket, the only bucket that still asks the compiler. |
 | `scripts/run-fail-sentence-test.mjs` | The wording of failure sentences that reach the user, in two parts. (1) A scenario refusal now has one sentence owned by the package: `scenarioNotAllowedContent` renders the subject, the allowed list (names escaped but never truncated, empty list omitted) and, for a fresh request only, the way out of omitting the scenario; the resumed-work form leaves that way out off, since a decision has no scenario to omit. The four decision exits (approve, deny, question card, plan review) append the sentence after the resume-policy sentence when the engine folded the refusal into `resume_blocked_by_policy`, and the failed outcomes of the two approval legs carry the folded original code as `blockedBy` next to the outer `errorCode` (present only when the outer code is the fold code and the body names a non-empty original). (2) The remaining user-facing failure sentences read a thrown value the same single way as the earlier seven: the four write and read exits (persisted rules, session policy, memory governance, MCP probe) no longer fall back to the class name or `[object Object]`; the subagent steer exit and the plan-review outcome sentence use the same reading and send the original thrown value's log form to the host log; the parked-row wait and the local rule parse take an optional `onFailure(log)` so their pure modules can hand the original to the host log; and the retry row's stalled `detail` goes through the same display exit as the four error arms. |
 | `scripts/run-engine-account-keys-projection-test.mjs` | The projection of five engine keys that older packages dropped on the way to the screen: a tool call's `recovered` start frame and `notExecuted` end frame (literal-true marks, read from own data properties only — never from a getter — and never turned into a claim about whether the call ran: absence means nothing), the completed run's machine-readable detail (the provider ended the last turn with a stop reason the engine does not read), the provider's account-condition word on a refused run, and the two notices that report an unknown stop reason and a stream that kept talking after its first ending. The suite drives the built package through the projection, the transcript adapter and the print-mode frame, and checks that absent, falsy, malformed and hostile values produce no claim and never throw, that the terminal reader keeps the run's kind whatever its detail looks like and isolates each key, and that the single reader for the marks is the only way the sources read them. |
@@ -552,10 +552,10 @@ guard still cross-checks the table by name).
 | `scripts/run-active-run-busy-read-projection-test.mjs` | How the package reads a rejected submission (the session already has a run in flight) in its terminal row, in its result frame and in the public discriminator `activeRunBusySignal`: one discriminator body and one carrier rule, shared by all three. The server's rejected and conflict terminal frames carry `status: 'failed'` since 7.125.0; the older idempotent-replay form — a cached copy of the original 409 conflict body wrapped as `done{ result, replay: true }`, with neither `status` nor a terminal word — is no longer recognised, since the package's support floor moved past the servers that sent it. The suite pins what such a body reads as today (the discriminator answers null, the row is the single `Outcome unknown` line, the result frame carries the unknown-outcome flag, the public projector agrees, with or without the replay flag, code only, id only), checks over a pool of sixteen frames that the discriminator and the row agree in both directions (including a cause-form frame that carries the busy code), pins which objects count as the carrier, pins that the retired folding function and its flag are gone from the source and the build, and pins in source that the discriminator and the carrier rule are each defined once. |
 | `scripts/run-plan-review-outcome-verdict-test.mjs` | The plan-review decision outcome has one reading rule: each outcome word is given only on positive evidence, otherwise `unconfirmed`. A decision reply that says the review is still pending (`retriable: true`) reads `still_parked` whatever the follow-up status read says; only `completed` reads `took_effect`; `suspended` reads `advanced` only when the follow-up read shows the new pause and that pause has a decision card (an approval or a plan review); `needs_review` reads `still_parked` when the review's instance identity before and after the decision can both be read and match, `advanced` when they differ and the new pause has a decision card, and `unconfirmed` otherwise; a malformed status on the follow-up read falls back to the decision reply like a failed read; every other status (`failed`, `blocked`, `timeout`, `running`, unknown words) reads `unconfirmed` and the sentence states the reading instead of claiming completion. `decidePlanReview` takes an optional `parkedResult` (the paused result the card answers) to supply the before-identity. Pinned with a structural injected client: failed approvals and rejections, the still-pending reply with and without a follow-up read, changed / same / missing / malformed identity, completed with and without the follow-up read, the other status words, the callback and the queue item carrying the same outcome word, and the new option on the shipped declaration. |
 | `scripts/run-durable-decide-leg-test.mjs` | Which durable decision leg a pending row belongs to has one table and one reading. The approval-card gate kinds are `human` and `irreversible_ask` (the only kinds the approval decide route accepts), the plan-review gate kind is `plan_review`, and the review wording set is `plan_review` and `needs_review`; each table is pinned at compile time in both directions against the engine's pause registry and checked again at run time against the installed engine. `durableDecideLegOf(row)` answers one of five words: `approval`, `plan_review`, `other_gate` (a known gate neither decide route accepts), `unreadable` (a gate word outside the registry) and `absent` (no readable gate word on the row: missing, null, empty, not a string, or a throwing read); it never throws. `askParkForeignGateKind` derives from it. Pinned with the twenty-shape verdict matrix, the derived foreign-kind answer, the row classifier, the pending-row lookup (including refusing to fall back to another row when the row a park names is not an approval), the busy-session recovery on non-gate words, the five-word union on the shipped declaration, and no engine-package reference in the shipped output. Since 0.97.0 the same table also decides which approval arm a parked frame goes to: `isToolApprovalGate` agrees cell by cell with `durableDecideLegOf` (an approval gate kind takes any tool that is not the question tool, an unreadable kind falls back to the file-write and shell name families, any other kind takes neither arm); a real-wire park on an approval gate kind for a tool outside those families raises the approval card and sends the decision end to end, while the non-gate word `tool_approval`, a plan-review kind with a write tool, and a non-gate word with the question tool raise nothing. |
-| `scripts/run-engine-caps-landing-projection-test.mjs` | How a client reads the whole capabilities answer of an engine, and how it swaps the probe used for later re-reads. The whole-answer reader has three states: landed (carrying the frozen snapshot that every narrow capability reader and every per-key reader also read), not landed (no probe yet, still in flight, failed, or invalidated), and unreadable (the probe returned a body whose root cannot be read). A key the engine did not report is unknown and never counts as on: the per-key readers are pinned to a written rule applied to the whole answer, and every narrow reader gives its own not-observed reading on an unreadable root. The same JSON parsed in another realm reads exactly like one parsed here: landing and the snapshot share one plain-record rule, and class instances stay non-records in both realms. A rotation entry for the same engine with a new probe (for example a client rebuilt after a credential refresh) starts that probe when nothing is landed or in flight, and otherwise only replaces the probe used by later single-flight re-reads without invalidating anything; the guard pins that the next re-read calls the new probe and never the old one, together with the in-flight, re-entrant and invalidation orderings. |
+| `scripts/run-engine-caps-landing-projection-test.mjs` | How a client reads the whole capabilities answer of an engine, and how it swaps the probe used for later re-reads. The whole-answer reader has three states: landed (carrying the frozen snapshot that every narrow capability reader and every per-key reader also read), not landed (no probe yet, still in flight, failed, or invalidated), and unreadable (the probe returned a body whose root cannot be read). A key the engine did not report is unknown and never counts as on: the per-key readers are pinned to a written rule applied to the whole answer, and every narrow reader gives its own not-observed reading on an unreadable root. The same JSON parsed in another realm reads exactly like one parsed here: landing and the snapshot share one plain-record rule, and class instances stay non-records in both realms. A rotation entry for the same engine with a new probe (for example a client rebuilt after a credential refresh) starts that probe when nothing is landed or in flight, and otherwise only replaces the probe used by later single-flight re-reads without invalidating anything; the guard pins that the next re-read calls the new probe and never the old one, together with the in-flight, re-entrant and invalidation orderings. It also pins the root-level nested-value reader for group capabilities (a group is either off or an object of flags), the change-subscription entry (one beat exactly when the whole-answer reader starts returning a different record, per base address, callbacks isolated per subscriber, no replay, idempotent unsubscribe, a seeded random operation sequence proving no extra and no missing beats), and the invalidate-and-reprobe entry that reuses the registered probe. |
 | `scripts/run-rule-tool-name-judge-test.mjs` | One reading for "does the engine accept this rule tool name": the engine's own tool-name check, installed as a value port and read through `ruleToolNameVerdictOf` (current / retired with its current name / unprefixed namespace / unmintable segment / unknown). The session-rule deliverability check asks it once for every name that survives the plain-text steps (an ordinary name the installed engine has newly retired is withheld as `legacy_tool_name`; a check that cannot answer makes the batch unreadable), the refused-entry removal removes every entry it does not call current and answers `judge_failed` without writing when it cannot answer, the sub-agent tool-list pre-check reports retired, unprefixed and unmintable names from it, reports an entry it cannot check as unjudged (sent as written with a warning, like any other deny-list problem) and judges case twins only against a supplied roster, and the request-body assembly rewrites a retired name to the current name it gives only in the deny and ask buckets (allow keeps it). No build-time retired-name table remains in the source. A real engine run that mounts no tool of that name, reading a stored session record, refuses to start exactly for the names this reading does not call current. |
-| `scripts/run-agents-wire-directory-test.mjs` | Agent tool lists carry what the agent definition says. Since 0.98.0 the supported floor's engine (server 7.129.1, engine 7.51.2) only warns about a name in an agent's tool list that nothing answers to, so the agents projection sends `tools` and `disallowedTools` as written — it no longer holds back an agent, drops an allow-list entry the engine's tool-name check flags, or drops an `mcp__` entry whose server the run does not declare — and the run-level exclude-list check `excludeToolsAgainstDirectory` is gone; whether a tool answers is answered by the engine alone. The tool-name check still reports its findings for both lists; an allow entry also gets a warning line, conditional on purpose ("gives the agent nothing unless this run has a tool with exactly that name" — when the run does have such a tool, mounted or in the delegated pool, the engine accepts the entry silently and it takes effect), while a deny entry gets no warning line at all, whether or not a JavaScript host still passes the old withhold switch. `engineToolDirectoryOf` still reads the MCP servers a run declares from two halves the host holds — the request's own `mcpServers` and the servers the deployment declares itself — minted into tool-name prefixes by the engine's own naming function; either half missing reads as not landed, an unreadable half or a missing naming port as unreadable, and an empty pair as an empty directory; the projection no longer reads it, so any directory, or none, yields the same payload. The gate checks every case against the engine judge of the supported floor (the installed engine package, which must be at or above the declared engine floor, and the live server fixture's engine when it is within the floor): every list sent as written must be accepted, with a warning where nothing answers, while an entry with a grammar flaw is still refused. 0.98.0: agent tool lists, names and maxTurns go on the wire as written (the live server fixture names the field), a written `tools` always yields `allowTools`, a retired deny entry is rewritten only when a roster is in hand and does not contain it, and the readiness gate returns an outcome (projection failure / not ready / ok) that recovers once the projection works again. |
-| `scripts/run-cloud-mcp-dropped-notice-test.mjs` | The sentences shown to a person when the cloud configuration leaves MCP server entries out. Each of the five reasons an entry is left out (switched off, a required setting missing, an unsupported transport, more entries than this client passes to the engine, both identity-header settings set) has its own sentence that names the entries; the switched-off sentence is informational and the other four are warnings. The sentences are part of the single wording source `cloudEffectiveNotices`, whose input now also carries the list of left-out entries. A reason outside the known set and an unreadable row (not an object, a reason that is not a string, a throwing getter, a reason or name that only exists on the prototype chain, an index that is an accessor or a hole) get a plain warning that names what can be read instead of being skipped, a list that is not an array or cannot be read (an accessor that is never run, a value of undefined, a proxy that throws) gets one warning saying it could not be read, a missing key adds nothing, and an empty list adds nothing. Names are capped at five with a `+N more` tail, repeated names are shown once, rows without a readable name are counted, and the order is degradation warnings, catalog references, left-out MCP entries (warnings first), budget, runtime caps. Pinned on the shipped declaration (the reason union is exactly the five words) and on the source (the sentence table is typed on that union and looks reasons up by own key). |
+| `scripts/run-agents-wire-directory-test.mjs` | Agent tool lists carry what the agent definition says. Since 0.98.0 the supported floor's engine (server 7.129.1, engine 7.51.2) only warns about a name in an agent's tool list that nothing answers to, so the agents projection sends `tools` and `disallowedTools` as written — it no longer holds back an agent, drops an allow-list entry the engine's tool-name check flags, or drops an `mcp__` entry whose server the run does not declare — and the run-level exclude-list check `excludeToolsAgainstDirectory` is gone; whether a tool answers is answered by the engine alone. The tool-name check still reports its findings for both lists; an allow entry also gets a warning line that promises no outcome ("an engine that refuses such an entry does not start the run (code config.agent.unknown_tool), and an engine that accepts it gives the agent no tool for this entry unless this run has a tool with exactly that name"), while a deny entry gets no warning line at all, whether or not a JavaScript host still passes the old withhold switch. `engineToolDirectoryOf` still reads the MCP servers a run declares from two halves the host holds — the request's own `mcpServers` and the servers the deployment declares itself — minted into tool-name prefixes by the engine's own naming function; either half missing reads as not landed, an unreadable half or a missing naming port as unreadable, and an empty pair as an empty directory; the projection no longer reads it, so any directory, or none, yields the same payload. The gate checks every case against the engine judge of the supported floor (the installed engine package, which must be at or above the declared engine floor, and the live server fixture's engine when it is within the floor): every list sent as written must be accepted, with a warning where nothing answers, while an entry with a grammar flaw is still refused. Since 0.100.0 the real-judge cells hold two expectation sets by engine generation: from engine 7.52.0 on, an entry that spells a mounted tool another way (a retired name that has a successor, a case twin) refuses the run before it starts with `config.agent.unknown_tool`, while earlier engines accept it and warn; both generations run on every fixture, and the warning line quotes the code the real engine throws. 0.98.0: agent tool lists, names and maxTurns go on the wire as written (the live server fixture names the field), a written `tools` always yields `allowTools`, a retired deny entry is rewritten only when a roster is in hand and does not contain it, and the readiness gate returns an outcome (projection failure / not ready / ok) that recovers once the projection works again. Since 0.99.0 the real-judge cells use the engine generation at the support floor; the newer installed engine is printed as a next-generation reading only. Since 0.100.0 the tool-directory reading has a fourth state — the request half read and the deployment half absent reads as `request_only`, carrying only the request half's prefixes and never posing as the complete set — and the warning sentences of the projection are disclosed when the payload is handed out rather than when a projection is accepted: a generation that was superseded before it was handed out says nothing, a sentence already said is not repeated across generations, and a refused-server naming input adds one sentence per allow-list entry that names a tool of a server the host left out of the run, byte-identical to not passing it when absent, empty or unusable. |
+| `scripts/run-cloud-mcp-dropped-notice-test.mjs` | The sentences shown to a person when the cloud configuration leaves MCP server entries out. Each of the eight reasons an entry is left out (switched off, a required setting missing, an unsupported transport, both identity-header settings set, an invalid `toolAxes` declaration, an `elicitation` value that is not a boolean, a name the settings contract's reader rejects, a transport the engine cannot use as written) has its own sentence that names the entries — the last two name a bad-name summary and a key path — and the switched-off sentence is informational while the other seven are warnings. An entry that stays on the wire without an optional part of its transport adds one warning that names the entry and the key path, never a value, and nothing is left out for exceeding 32 entries any more. The sentences are part of the single wording source `cloudEffectiveNotices`, whose input now also carries the list of left-out entries. A reason outside the known set and an unreadable row (not an object, a reason that is not a string, a throwing getter, a reason or name that only exists on the prototype chain, an index that is an accessor or a hole) get a plain warning that names what can be read instead of being skipped, a list that is not an array or cannot be read (an accessor that is never run, a value of undefined, a proxy that throws) gets one warning saying it could not be read, a missing key adds nothing, and an empty list adds nothing. Names are capped at five with a `+N more` tail, repeated names are shown once, rows without a readable name are counted, and the order is degradation warnings, catalog references, left-out MCP entries (warnings first), budget, runtime caps. Pinned on the shipped declaration (the reason union is exactly the eight words) and on the source (the sentence table is typed on that union and looks reasons up by own key). |
 | `scripts/run-tool-card-outcome-test.mjs` | One reading, one sentence, for what a tool result card in the transcript actually says. Three marks decide it: the placeholder mark on the record (the engine sent no body, so the content is an empty placeholder), the engine's positive evidence that the call never ran, and the block's error flag. `readToolCardOutcome(record)` reads them as `normal` (a real result; success or failure still comes from the error flag and is never rewritten), `degraded` with a cause and one sentence (`not-executed` — takes precedence and is independent of the error flag, so a card that looks successful but never ran is caught; `error-without-detail`; `no-result`, where the error flag false is a placeholder and not a success; `unrecognised`, a placeholder value from another version) or `unreadable` (not a tool result record, or the placeholder position cannot be read — never read as normal). The placeholder has two minting points that cannot be told apart on the record (a call closed at the end of the turn with no end frame, and a real end frame without a body), so the no-result sentence states only what both share. Pinned on records produced by the real pipeline in both lanes (the transcript row and the print-lane frame answer the same), a read-failure matrix that never throws and never reads as normal (accessors are not run, prototype-chain keys do not count), four distinct sentences with no internal names or tool-family names, and the source (the sentence table is typed on the cause union, the evidence mark goes through the existing strict-true reader, the placeholder constant is imported not re-written, and each sentence has exactly one minting site in the shipped output). |
 | `scripts/run-approval-settled-single-source-test.mjs` | Whether an approval already has an outcome is recorded once per process, in the package's approval-holder registry, at two points only: when the live card's holder settles (the decision was delivered and acknowledged, or the interruption denial it sent was accepted) and when the engine's close-out frame for that approval arrives. Every path that could put up a card for the same approval reads that one record first: the in-stream frame path, the suspended-ask path (same entry point), and a host's own durable route through `joinApprovalHolder`, which now answers `joined` for an approval that was already answered or closed by the engine even after the holder let go (an approval the engine moved to the parked queue still answers `free`, since it remains open there). Cells cover the host's durable route after a delivered answer and after a close-out with no card ever shown, the parked negative control, the suspended-ask path writing and reading the same record, a new bridge still recognising a settled approval, the replay verdict reading only this record (a host-supplied map is ignored), the outcomes that are not recorded (a delivery failure including 404, a withdrawn card, a refused edit, an interruption denial that failed, threw or was answered with the parking reply, or that a parked close-out overtook), the write happening at the acknowledgement rather than when the card is answered, and the 4096-entry bound |
 | `scripts/run-server-7126-pick-test.mjs` | Checks for the 7.126.0 server and SDK 21.1 pickup. A submission refused because its idempotency key was already bound to a different request body is classified by one code check on every leg (thrown errors and the refusal in the stream's final frame), and the result frame carries the package's own sentence. The engine's approval card word is read through one closed-set reader and reaches the approval card from both the live frame and the pending row; a missing or unknown word stays missing. The run's failure detail is read as the server's three-arm union plus an unrecognized arm. The two decision doors' gate-kind tables are the SDK's own objects, and the retired hand-written copy is gone from the root entry. The retired team-review request keys are refused at assembly, and allowTools is sent as given. The models list's rejected-candidate record has a three-state reader. Against the 7.126.0 server package it checks the request key closure, the 409 mint site, the engine's card words and the failure-detail registry. |
@@ -564,35 +564,61 @@ guard still cross-checks the table by name).
 | `scripts/run-engine-url-transport-test.mjs` | Engine address handling in one place. `normalizeEngineBaseUrl` trims outer whitespace, strips every trailing slash and keeps the user's spelling; it refuses, with one of five closed reasons (`empty`, `scheme`, `unparseable`, `userinfo`, `query_or_fragment`), anything that is not a plain http(s) base address. A query or fragment is refused even when empty, because request paths are appended to the base as plain text and would end up inside the query. `engineUrlTransportVerdict` lets `https:` through and `http:` only for a host the settings contract's narrow local-host check accepts, the same check that decides whether outbound credentials are dropped, so the credential fence and the plaintext gate cannot disagree about one host; anything else is refused as plaintext to a remote host or as not an http(s) address, and an address it cannot read is refused. Neither function returns message text or echoes the input. The gate checks every refusal reason and success form, the two addresses the terminal and the desktop shell used to answer in opposite ways, that the local-host reading has a single call site over the settings contract's own function and never touches the wider routing check, that the verdict matches the installed settings contract on a host table, that hostile input never throws and a very long run of slashes stays linear, and it compares the package with copies of both shells' former logic on a generated corpus: the package accepts exactly what both former implementations accepted, with the only disagreements between them falling into two registered classes. The cloud configuration's session-mirror address is read through the same normaliser: well-formed addresses come out byte-identical to the former trailing-slash trim, an address the normaliser refuses is passed on as read with its `required` flag intact rather than dropping the mirror, and a long run of slashes stays linear. |
 | `scripts/run-submit-reopen-test.mjs` | A streaming submit whose connection closes before the run is bound is reopened exactly once with the same idempotency key. `submitStreamWithBoundedReopen` calls the SDK's streaming submit itself, so the key it mints and the response hook that reads the run id header cannot be forgotten by the host; the host passes the request and its own stream options (its own response hook is called after the package's). It reopens only when every one of these holds, and anything it cannot read counts as a reason not to reopen: no run id header was seen (an unreadable header counts as seen) and no frame was delivered; the caller has not cancelled (an unreadable signal counts as cancelled) and the consumer has not asked the stream to stop; the engine did not answer with a status code; the transport-failure reading is readable; and the failure is a lost connection with no answer according to the package's own reading. Cancellation is checked again after the failure is classified, just before the second send. A second failure is thrown as it is, so there is never a third send, and a close after the run is bound is left to the run-event continuation. A reading reports the key, how many sends were made, and whether the terminal frame was a replay of the earlier request under the same key. The gate drives the real SDK stream through a recording fake fetch and checks the key header on every POST, the POST count (also with an SDK client configured to retry, since its streaming leg is single-shot), each case that must not reopen, the replay reading from both of its sources, and that a consumer stopping the stream before its first frame prevents the second send. |
 | `scripts/run-protocol-name-reads-test.mjs` | Reads of sdk error class names when the machine code is gone, checked in real bundles. The sdk gives every error instance a name taken from its class, and a production bundle's minifier renames the class, so comparing that name to a literal never matches there. The judges now treat a class name as family evidence only on rebuilt errors (plain objects that kept just a `name`), compared against one closed table of literal protocol names through one reader; a real sdk instance is judged by its machine code or status, never by its name. The gate bundles the sdk errors and this package three ways (plain, minified, minified with kept names) and runs real instances, literal rebuilt forms and real instances with the code stripped through eight judge outputs, expecting identical answers in all three bundles; it also pins that the literal names live in one place in the shipped output, that they equal the installed sdk class names, the three-state name reader, and the order in which code and name are consulted. |
-| `scripts/run-thrown-data-reads-test.mjs` | Machine-read slots on thrown errors, read the same way everywhere: only an own data property counts. Four readers (the effective code behind a folded refusal, the scenario refusal reader, the blocked-by reader and the deny-attribution refusal reader) and two siblings used to read a status or code along the prototype chain and run accessors. The gate drives every reader with each trigger shape and each slot placed five ways (own data, prototype chain, accessor returning, accessor throwing, a proxy whose descriptor trap throws) and expects the answer for the slot being absent (an explicit unreadable state for the blocked-by reader), no accessor ever run and no reader ever throwing; real sdk errors read exactly as before, and a syntax-tree pin keeps ordinary property reads of those slots out of the four function bodies. |
+| `scripts/run-thrown-data-reads-test.mjs` | Machine-read slots on thrown errors, read the same way everywhere: only an own data property counts. Four readers (the effective code behind a folded refusal, the scenario refusal reader, the blocked-by reader and the deny-attribution refusal reader) and two siblings used to read a status or code along the prototype chain and run accessors. The gate drives every reader with each trigger shape and each slot placed five ways (own data, prototype chain, accessor returning, accessor throwing, a proxy whose descriptor trap throws) and expects the answer for the slot being absent (an explicit unreadable state for the blocked-by reader), no accessor ever run and no reader ever throwing; real sdk errors read exactly as before, and a syntax-tree pin keeps ordinary property reads of those slots out of the four function bodies. The generated pool now covers fourteen failure readers plus three private classifiers driven through public entries, and a new section pins the not-found reader that tells a typed not-found error, a not-found machine code and a code-less 404 apart. |
 | `scripts/run-turn-failure-presentation-test.mjs` | The failure sentence entry point, the turn failure presentation and the reattach budget. `wireFailureTextsOf` turns anything thrown into a sentence that is safe to show (never empty, never a class name or an object tag, never throwing) plus a log form that is not for the screen. `turnFailurePresentation` classifies a failed turn from wire facts only: ten categories, whether to offer a retry (the server's own retriable flag first), how long to wait (only when the server gave a window, rounded up, never zero), and a host-neutral sentence for the failures that carry no engine message. `runReattachPolicy` decides whether and when to re-attach a run whose stream ended, spending one explicit budget. The gate pins each absence shape, a table of real sdk errors, the own-data-slot reading, and compares the budget against reference copies of the web and desktop loops over a thousand generated sequences. 0.98.0: the seat `error` event superset key is minted only by `seatTurnFailureOf` (field-for-field the turn-failure presentation) and read only through `readSeatTurnFailure`, which validates its shape and returns undefined for anything malformed. |
 | `scripts/run-sandbox-hint-host-test.mjs` | Sandbox degradation hints per host. Each of the five hints is a fact sentence (what the engine refused and why, the same for every host) followed by a way out that depends on the host: a terminal can re-run without the sandbox flag or list profiles with a command, while a graphical host has neither and points at keeping the default environment or choosing another one. With no host given the hint is byte-for-byte what it was before; an unrecognised host gets the fact sentence only. The gate checks the terminal wording against the previous release verbatim, that graphical hosts get the fact sentence and no terminal-only words, that detection does not depend on the host, and that status and code are read as own data properties without ever throwing. |
 | `scripts/run-transcript-identity-projection-test.mjs` | Transcript identity: every id this package hands out for a transcript row, a projected frame, a segment, an attachment event or a per-response message id is a deterministic function of the engine identity it carries, so projecting the same wire events again (a second process, a reconnect replay without a cursor, a resume replay, a desktop replay) yields the same ids and a uuid-keyed transcript gets no duplicates. Driven through the real pipeline with two independent adapter instances and random sources: live and durable event sets each projected twice give equal row uuids, segment ids, segment-end committed uuids, per-response ids and engine references for every row class; a projection joined mid-stream matches the full one for every row whose opening event it saw; a streamed text or thinking row is identified by its first and last contributing events, so two projections that cut a segment at different points (an idle split, a stream ending mid-segment, a rewritten authoritative segment) never share a row id and a uuid-keyed transcript keeps both instead of losing text, while a child process projecting the same events yields the same ids; the `_sema_engine_ref` reference on frames, rows and print-mode tool result frames carries the opening event id plus the run binding only when the host supplies it, and is absent for host rows, old rows and events without identity; `sameEngineRef` answers same / different / unknown and can only answer same when both sides carry an event id; a frame carrying only a stream cursor (`seq`) no longer gets a derived id; terminal rows and result frames are derived from the run id when present; a print-mode tool result frame and the interactive tool result row share one id, the engine event id. An optional segment compares against the previous release build and fails on any unregistered id change or any change to row count, order, class or body. |
-| `scripts/run-inherit-env-intent-projection-test.mjs` | Whether the interactive lane may ask for the user's whole environment in tool shells. The engine says, through one capability word on its capabilities answer, that it keeps its own credentials out of shells it starts; this guard pins the four readings of that word (confirmed only for a literal top-level true; absent, declined with the value's shape, unreadable, or not yet observed otherwise, never read from a getter), the one decision built on it (interactive lane, project context confirmed, word confirmed and identity verified, all four, or the request omits the field with one named reason), and the single sentence clients show when the word is not confirmed. Looser spellings, nested groups and stringly true values must never open it, and the server fixture is checked by generation: one before 7.129.0 does not publish the word anywhere in its release, while 7.129.0 and later mint it at the top level of the capabilities answer, and the real minting function's answer, landed through the capabilities cache, is read by this reader as confirmed or declined by its boolean value. |
+| `scripts/run-inherit-env-intent-projection-test.mjs` | Whether the interactive lane may ask for the user's whole environment in tool shells. The engine says, through one capability word on its capabilities answer, that it keeps its own credentials out of shells it starts; this guard pins the four readings of that word (confirmed only for a literal top-level true; absent, declined with the value's shape, unreadable, or not yet observed otherwise, never read from a getter), the one decision built on it (interactive lane, project context confirmed, word confirmed and identity verified, all four, or the request omits the field with one named reason), and the single sentence clients show when the word is not confirmed. Looser spellings, nested groups and stringly true values must never open it, and the server fixture is checked by generation: one before 7.129.0 does not publish the word anywhere in its release, while 7.129.0 and later mint it at the top level of the capabilities answer, and the real minting function's answer, landed through the capabilities cache, is read by this reader as confirmed or declined by its boolean value. Since 0.100.0 it also pins the sibling word for credential values (a fact that can be false; absent on older servers) read the same way by its own reader, which the intent judgement does not consult. |
 | `scripts/run-approval-hold-scope-test.mjs` | One holder per approval across every card-opening path. The live-card entry, the parked-row decision path and the host's own paths (registered through `claimApprovalCard`) all register in one table and see each other through one symmetric matching rule: same approval id when both sides have one, otherwise the same run and the same call while the holder is still waiting for the human. A later arrival waits for the holder and, when the holder settled the question, sends nothing and gets the holder's outcome. Also checks the superseding-card position (`supersedes`, with the earlier answer disclosed only when the engine did not take it), the parked-row position (`parkRow`, deadline from the row's own SLA wall) and the parking notice function. |
 | `scripts/run-hitl-answer-outlets-projection-test.mjs` | The new decision-chain outlets. `answerPayloadFor` builds the base decision body for both lanes from the one answer object (decision word, session approval, edited input, the deny note behind its capability and length gates) and the package's own lanes use it; a Yes note never reaches the wire while `answerCarriage` answers `none`; the answer-committed event fires exactly once per answered card before the first send. `rememberOfferable` is checked card by card against `standingRowsOf`; `localAllowRuleEnforceable` against the permissions declaration judge. `lateDecideOutcomeOf` is checked against a frozen copy of the previous host-side classification. `RESPOND_DECISIONS` is frozen with exactly three words. |
 | `scripts/run-seat-apply-config-reason-test.mjs` | The seat's model-configuration receipt now carries a closed set of reasons instead of a free-form string. When a host refuses to write the configuration it answers with one of five words — the model catalog on this device shadows the write, the cheap model asks for a different provider, the request was not valid, the file could not be written, or the app cannot change the configuration right now — and the one-line human sentence for each word is minted in exactly one place, so a client renders it instead of keeping its own table. Free-text diagnostics moved to an optional `detail` field that is display-only. The outbound gate refuses any other word, including prototype-chain names and case variants, and refuses a failed receipt with no reason. The guard also reads the built declaration for the exact member list, compiles probes that must fail to type-check, and pushes each word through the gate, a structured clone and the sentence reader to show the renderer reads the same sentence the host chose. |
-| `scripts/run-seat-stream-history-test.mjs` | The seat's streaming-preview chunk and the engine session history projection each have one definition now. Live preview text used to be a shape the desktop host built by hand and the web client unpicked by hand; the package now mints it (`seatStreamDeltaMessage`) and reads it (`readSeatStreamDelta`, which tells a chunk, an unknown inner event, a malformed chunk and an ordinary message apart, so a chunk is never mistaken for a committed message). Rebuilding a transcript from the engine's stored session rows used to be written once per client with diverging rules; `projectEngineSessionMessages` is the single projection: it cuts out the engine's own prepended content using the provenance marks the engine stamps on each user row (never by guessing from text shape, so a user's own text that merely looks like a reminder tag is kept), keeps tool results as tool-result blocks, maps the engine's block vocabulary to the chat-message vocabulary, never invents a timestamp the row does not carry, copies tool arguments and unknown blocks as plain JSON so what it returns can always be cloned and sent, and reports every row or block it could not read instead of dropping it silently. The guard checks the minted bytes against the previous hand-built shape, runs a real adapter stream through mint, gate, structured clone and read, pins ten stored rows to an exact expected transcript, checks the provenance rules case by case (segments across blocks, malformed ranges, absent marks), and feeds hostile rows (getters, throwing proxies, sparse arrays, functions and cycles inside tool arguments) to show one bad row never takes the others down. |
+| `scripts/run-seat-stream-history-test.mjs` | The seat's streaming-preview chunk and the engine session history projection each have one definition now. Live preview text used to be a shape the desktop host built by hand and the web client unpicked by hand; the package now mints it (`seatStreamDeltaMessage`) and reads it (`readSeatStreamDelta`, which tells a chunk, an unknown inner event, a malformed chunk and an ordinary message apart, so a chunk is never mistaken for a committed message). Rebuilding a transcript from the engine's stored session rows used to be written once per client with diverging rules; `projectEngineSessionMessages` is the single projection: it cuts out the engine's own prepended content using the provenance marks the engine stamps on each user row (never by guessing from text shape, so a user's own text that merely looks like a reminder tag is kept), keeps tool results as tool-result blocks, maps the engine's block vocabulary to the chat-message vocabulary, never invents a timestamp the row does not carry, copies tool arguments and unknown blocks as plain JSON so what it returns can always be cloned and sent, and reports every row or block it could not read instead of dropping it silently. The guard checks the minted bytes against the previous hand-built shape, runs a real adapter stream through mint, gate, structured clone and read, pins ten stored rows to an exact expected transcript, checks the provenance rules case by case (segments across blocks, malformed ranges, absent marks), and feeds hostile rows (getters, throwing proxies, sparse arrays, functions and cycles inside tool arguments) to show one bad row never takes the others down. A replayed tool-result row that the engine archived as parked at an approval gate now carries a machine-readable parked flag on its tool-result block, so a replay does not render it as a failure. |
 | `scripts/run-assistant-inbox-projection-test.mjs` | One reading of the per-user inbox of parked questions. `readAssistantInbox(assistant)` returns `present` (the rows in exactly the order the server sent them — the package neither sorts nor re-sorts; a missing severity stays missing, it is not a lowest tier and not zero; marks that are only ever sent as true are read as true only, so an absent bidi mark is never "checked clean"; malformed rows are dropped and counted; an empty inbox is a real empty list), `unavailable` (unauthorized; unreadable — a body without the array is never rendered as an empty inbox; or failed, with the status and code) or `not-reported` (the route itself answered 501, or no assistant connection was handed in — the whole panel stays hidden). Whether the lane exists is asked of the call's own answer, never of a capability flag, and the background view's assistant leg uses the same judge. Pinned on the real client against stubbed fetches, on fixed order cases with a syntax-tree scan that the source never sorts, and against the fixture server's key list and sort law. |
 | `scripts/run-control-preempt-test.mjs` | Graceful yield for a running run. `ControlRouter.preempt()` asks the engine to pause the run at its next clean turn boundary (the run can be resumed later), unlike cancel, which kills it. A 202 is not a success by itself: the answer's status word decides — preempting means asked; running means this run cannot yield and nothing changed. At most once, never re-sent; failures use the same judge as steer and cancel. When the engine's yield route is unavailable (it answers 501, or there is no assistant connection) the default falls back to cancel — a hard stop, reported as such — and the host can instead choose to just report it. Each outcome has one neutral sentence, and none says the run has yielded. Pinned on the real client (even with retries enabled there is exactly one send), on a pool of thrown values judged by the shared at-most-once judge, on import-cycle smoke runs and on the fixture server's minting points. |
 | `scripts/run-device-registration-test.mjs` | Device registration verbs and the identity-mismatch refusal. `issueEnrollToken`, `enrollDevice` and `revokeDevice` go through the client's devices slice with the same closed refusal table and at-most-once judge as the rebind calls, and every outcome is a value, never a throw. The one-time enrolment token appears only in the issue call's ok outcome — never in another outcome, in a sentence, or in a log call (a refusal sentence that echoes the token is dropped from the outcome). The enrol body is exactly the six closed keys in fixed order and nothing else the caller passed; a missing token, public key, platform or workspace root is refused locally with no request; the returned device id is accepted only if the host's own shape judge says so. A revoke is ok only if the answer echoes the same device id and the revoked status. The identity-mismatch refusal has one reader for its two forms (the machine code, and the bracketed prefix in a tool error text), and the bound device id is read only from structured fields, never from prose. Pinned on the real client with stubbed fetches (writes are sent exactly once even with retries enabled), on a plain-object pool of thrown values, on a syntax-tree scan, and on the fixture server's minting points. |
 | `scripts/run-durable-seq-reading-test.mjs` | One reading of a frame's resume position. `durableSeqOf(frame)` returns the cursor value of an event: the stream id comes first (a string or a number; a malformed id hides a valid sequence number, so the cursor does not move), and only when there is no id does a numeric sequence number count — except on progress, notification and task-start events, whose sequence number is a different axis (a sub-agent generation or a stop-cycle batch counter) and never a cursor; a frame with neither stays at the previous cursor. The host app's own copy of this reading answers differently on nine shapes; the guard freezes that copy and pins the exact disagreement set, extracts the real function from the host app tree when it is present, and pins that the gate ledger's cursor is fed by this one function only. |
 | `scripts/run-task-notification-xml-read-test.mjs` | The single-notification read for transcript task-notification XML. The package already had a batch reader that turns a saved transcript into de-duplication keys; a host that needed the task id and status of one queued notification wrote its own scan. Both now read through one block parser: a block is a `<task-notification …>…</task-notification>` pair, the id and status are the first `<task-id>` / `<status>` pair inside it, un-escaped, and a field that is missing, empty or unterminated is simply absent. The parser is linear: unterminated openers, openers with whitespace and no closing bracket, and look-alike tags no longer cost time proportional to the square of the input. The gate pins the single read against the render function over a generated pool (with escapes, whitespace and newlines), compares the batch reader with a frozen copy of the previous regular-expression implementation over thirty thousand generated texts in three message shapes, runs four adversarial inputs of hundreds of thousands of candidates in a child process with a timeout, and checks by syntax tree that neither entry point contains a regular expression of its own. |
 | `scripts/run-narrow-readers-export-test.mjs` | Four narrow readers for foreign objects, exported from the package root. Hosts used to write their own copies: a three-field reader for thrown errors (status, error code, message) existed three times with identical text, and the one-line checks for a non-empty string or a finite number appeared three to eight times per host. The package already had one reading rule for this, reading only the object's own data properties without running accessors, but it was internal. The four exports are the own-data string reader, the own-data finite-number reader (zero is a value, NaN and infinities are absent), a guard that turns a throwing read into an absent one, and a three-field reader for thrown values whose message is never empty and is produced by the same function as the user-facing failure sentence. The gate drives each reader with every absence shape (missing, inherited, accessor, proxy trap, wrong type), compares the three-field reader with a frozen copy of the host-side code over six thousand generated JSON-shaped errors (the only deliberate difference is the sentence used when no message exists), checks by syntax tree that the three-field reader is a projection of the existing snapshot reader, and compiles a small consumer under the strict settings to prove that an `unknown` from a catch clause can be passed straight in. |
-| `scripts/run-plan-review-offered-mode-test.mjs` | The offered-mode read for plan-review cards, and the retirement of the synchronous reopen overload. A plan-review card is shown either as the old two-choice card or as a three-choice card whose two Yes options each carry a mode to switch to after approval. Reading the user's answer correctly needs the labels that were actually shown, otherwise a three-choice Yes is read as dismissed. The package knew what it had shown but cleared it on the first answer and exposed no reader, so a host kept its own table of what it had armed and re-derived the rules around it (reopened cards are always two-choice, a missing entry means two-choice, an armed card is not rewritten by a replay). One reader now answers `three_choice`, `two_choice` or `unknown` from positive evidence only: the labels on the card or frame when given, then the question id (reopened identities are two-choice by construction, otherwise a bounded record the package keeps when it publishes the card, not cleared by an answer). `unknown` is never read as two-choice by the reader itself. Reopening a card now always returns a promise and requires the presentation-receipt window; a call without it presents nothing and resolves as not reopened. The gate drives the real arm, answer, dismiss and reopen paths, compares the reader with a frozen model of the host-side table over a hundred and sixty random steps across three tasks, and pins by syntax tree and a strict compile that the overload is gone. |
+| `scripts/run-plan-review-offered-mode-test.mjs` | The offered-mode read for plan-review cards, and the retirement of the synchronous reopen overload. A plan-review card is shown either as the old two-choice card or as a three-choice card whose two Yes options each carry a mode to switch to after approval. Reading the user's answer correctly needs the labels that were actually shown, otherwise a three-choice Yes is read as dismissed. The package knew what it had shown but cleared it on the first answer and exposed no reader, so a host kept its own table of what it had armed and re-derived the rules around it (reopened cards are always two-choice, a missing entry means two-choice, an armed card is not rewritten by a replay). One reader now answers `three_choice`, `two_choice` or `unknown` from positive evidence only: the labels on the card or frame when given, then the question id (reopened identities are two-choice by construction, otherwise a bounded record the package keeps when it publishes the card, not cleared by an answer). `unknown` is never read as two-choice by the reader itself. Reopening a card now always returns a promise and requires the presentation-receipt window; a call without it presents nothing and resolves as not reopened. The gate drives the real arm, answer, dismiss and reopen paths, compares the reader with a frozen model of the host-side table over a hundred and sixty random steps across three tasks, and pins by syntax tree and a strict compile that the overload is gone. It also pins the public question-id reader (task id plus a reopened flag, undefined for anything that is not a plan-review card). |
 | `scripts/run-context-usage-projection-test.mjs` | The one narrow reader for the engine's context-usage payload. The payload (used tokens, window, compaction threshold, and the prompt's section rows) was read in several places: the projection filtered section rows, but the next stage passed any array through, so a hand-built frame fed straight to the adapter put injected keys and invalid rows on the event; the terminal re-implemented the finite-number checks on raw frames to serve both the interactive and the print lane; the web client re-narrowed the rows on the already-projected event. The reader accepts the raw frame, the internal arm or the chrome event, because the four fields have the same names and shapes in all three, and returns each field independently: numbers must be finite (absent is never turned into zero), rows need a non-empty id and a finite token count, rows that fail are dropped one by one, an all-bad list leaves the key absent, and a payload with nothing readable returns nothing. The projection and the chrome arm both call it. The gate checks the table of shapes, accessors and proxies, compares the reader with a frozen copy of the web client's reader over eight thousand generated payloads, runs the same frames through the projection and straight through the adapter to prove the two entries agree, and checks by syntax tree that neither entry narrows rows itself. |
 | `scripts/run-tool-gate-view-projection-test.mjs` | The gate-record view on tool-result rows. The engine's tool_end frame carries a gate record (the disposition: allowed, or denied by which layer; how the wait was settled; who asked), and the package has one reader and a set of judges for it, but the projection only folded it into a seven-word category on the transcript row, so the ten denial layers and the settlement words stopped before the card and a host had to intercept the raw frame ahead of the adapter to read them. The tool-result row, in both the interactive transcript and the print-mode frame, now carries the same view under one envelope key, present only when the reader can read it and never inside the content block that is sent back to the provider. Absence says nothing: a missing view does not mean allowed. The gate runs five gate records through the real projection and adapter and compares the key with the reader's own result, checks the absent shapes (no record, malformed, unreadable disposition, parked), the print-mode frame, hand-built arms in both view and raw shape, idempotence of the reader over six hundred generated records, and that the five public judges answer identically on the row's view and on the raw frame's view. |
 | `scripts/run-connector-status-projection-test.mjs` | Reader-minted connector status. `connectorStatusOf({ name, mountFace })` reads one answer of `readMcpCapabilities` (this deployment's default MCP server list, each server dialed live) through the same per-name row lookup the MCP detail card uses: a fully read list without the name means unmounted; exactly one row whose engine reachability observation is unreachable means unreachable; exactly one reachable row means the connector speaks for itself (its own authorization state is not read here); everything else is unknown, with the step where the reading stopped. The dial verdict on the row is never read, because a failed dial includes servers that did answer (a protocol error, a 401). Pinned: the two states and their shape against the real bytes of the contract package's status schema, both directions; every combination of list shapes, reachability shapes and dial verdicts against an independent oracle; an unreadable list or name never becomes unmounted (the guard proves it catches a reader that folds them); per-key throwing inputs read once each; the ten distinct English sentences of `connectorStatusLine`, the same for every language tag; an end-to-end run through the real client, including that a caller-probe answer is never read as this deployment's list (it is rejected at compile time and reads as unknown at run time); a single minting point in source; and the supported server's and engine's minting points. |
 | `scripts/run-hitl-single-reading-projection-test.mjs` | One reading per fact on the approval surface. When a park carries the identity of the call it holds, the collateral marking of aborted sibling calls reads that identity directly, so two calls of the same family are told apart, while frames from a sub-agent lane or naming a different gate are left unmarked; without the identity the earlier inference is unchanged. The session-approval fallback re-sends a plain approval only for the dedicated refusal codes. Reopening a plan-review card with the reused question id and an injected delivery path cannot be expressed in types and is refused at runtime. The suspended-ask tracker reports only appeared and gone. The sub-agent test reads the explicit marker only. An unreadable pending section escapes into the poll-leg breaker while a partly readable one commits with a dropped count and still owes a complete read. Tool-start argument snapshots live in one table. A panel row keeps its first binding. The fleet description filter no longer hides tools named Task or Fork. The gate record view carries which seats were present but unreadable, so the denial-kind reader answers the same on the view as on the raw frame. Both tool-output text readers share one reading. |
 | `scripts/run-row-sidecar-projection-test.mjs` | Row sidecar facts. A terminal that the server marks as an idempotent replay (this request did not start a new run) carries `_sema_replay: true` on the result frame, on synthesized terminal rows and on rows derived from them, read through one guarded reader; `isReplayMarked` reads the fact; exactly one required `result_replayed` chrome event carries the user-facing sentence per replayed terminal, whatever the row count. Tool result rows carry four more facts on the envelope (collateral abort, never started, approval window expired, denial-limit window expired), minted by one function shared by the interactive transcript and the print frame and read with `toolResultRowFlagsOf`; `TRANSCRIPT_ROW_SIDECAR_KEYS` lists every row key a host must keep when it rebuilds rows. The gate pins the readers by type checker and syntax tree, the key order, the one-event count, the sentence, the consumer contract, the shared mint and the identity range. |
-| `scripts/run-tool-row-terminal-projection-test.mjs` | A tool call held by a park gets no terminal row from its park-time close frame: the call is not settled yet, so the projection takes its card off the open-card table, records it on the adapter instance (with the parked run and checkpoint id, delivered by an internal terminal-fact arm that the stream emits before the result frame) and emits one `tool_parked` chrome event naming the call and the transcript uuid of its start row. Its terminal row is minted only by the executing leg: a later frame for the same call id claims the original card only with a witness (the host declaring `continuesPark` for the stream it is projecting, or a continuation the question bridge spliced into the same projection); without one it is a new call with its own start row. A claim also needs the parked start row to still be the latest start row the instance minted for that call id (whichever of the two was recorded first), refuses a frame naming a different tool or bound to a run outside the park and its declared continuation, and keeps the original row identity while taking the continuation's input. Collateral calls, unreadable frames and a park the package itself settled (Esc on a parked card) keep their terminal rows; the print lane uses the same judge; the registry travels with `exportLedger` / `importLedger`; a run that ends without parking retires its entries; and results pair with the nearest preceding unpaired start row of the same call id. |
+| `scripts/run-tool-row-terminal-projection-test.mjs` | A tool call held by a park gets no terminal row from its park-time close frame: the call is not settled yet, so the projection takes its card off the open-card table, records it on the adapter instance (with the parked run and checkpoint id, delivered by an internal terminal-fact arm that the stream emits before the result frame) and emits one `tool_parked` chrome event naming the call and the transcript uuid of its start row. Its terminal row is minted only by the executing leg: a later frame for the same call id claims the original card only with a witness (a continuation the question bridge spliced into the same projection); without one it is a new call with its own start row. A claim also needs the parked start row to still be the latest start row the instance minted for that call id (whichever of the two was recorded first), refuses a frame naming a different tool or bound to a run outside the park, and keeps the original row identity while taking the continuation's input. Collateral calls, unreadable frames and a park the package itself settled (Esc on a parked card) keep their terminal rows; the print lane uses the same judge; the registry travels with `exportLedger` / `importLedger`; a run that ends without parking retires its entries; and results pair with the nearest preceding unpaired start row of the same call id. The host-declared continuation witness no longer exists: only a continuation spliced into the same projection reclaims the parked row, and across projections every continuation is a new call until the engine marks the frame itself. |
 | `scripts/run-session-policy-grammar-test.mjs` | Reading and rewriting a session’s own rule record never changes which rule grammar its entries are read in. The record carries a grammar stamp: the read exposes it (absent means a record stored before the stamp, which the engine reads in its earlier grammar — never folded into a number), and both rewriting paths — tightening and removing refused entries — send the same stamp back. When a record has no stamp and a rewrite would carry back one of its allow entries written in a tool’s own rule form (such as Bash(git:*)), which the earlier grammar reads as allowing nothing and the current grammar reads as a working allow rule, the rewrite sends nothing and returns a machine-readable refusal naming those entries — for every caller identity alike, since an identity allowed to widen rules would otherwise have the write accepted. An engine refusal of the same kind maps to the same outcome, and a write receipt that comes back without the stamp it was sent with, or with an older one, is never reported as written. The guard runs each case for both caller identities against a modelled engine and, when the matching server build is installed, against the real engine write judge and the server’s error mapping, recording the pre-fix widening as its failing baseline; over a generated corpus it checks that every entry the engine reads differently in the earlier grammar is one the package refuses to carry back, and that every extra entry it refuses is one the engine would refuse to write anyway. Malformed stamps, accessors, per-key throwing proxies and a polluted prototype are refused rather than read as absent. |
 | `scripts/run-message-catalog-test.mjs` | The package's message catalog. Every user-facing sentence the package mints is meant to live in exactly one place, a key in the catalog: `EN_MESSAGES` holds the English text and defines the key set, `ZH_MESSAGES` holds the Chinese text, and both tables are frozen. `translate`, `createTranslator`, `msg` and `renderMessage` pick one language per call; a host only chooses the language, and `resolveUiLanguageFrom({ saved, hostTags })` resolves it from a saved choice, then the host's ordered locale tags (the first non-empty tag decides, including `C`), then English. The gate checks that both languages have the same keys and the same placeholders, that no value is empty or left untranslated, the key grammar and domains, that no key collides with an upstream code, that backticked commands and registered terms are kept verbatim, that every value survives the display sanitizer, that the keys taken over from the terminal are byte-identical to a pinned snapshot, that the tables are frozen, that translation never throws (unknown language, inherited names, missing keys, missing parameters, a failing miss ledger), that nested references render in the outer language and are recognised across two copies of the package, and that the language resolver matches the terminal's previous resolution chain on six thousand generated inputs. |
 | `scripts/run-bare-sentence-ratchet-test.mjs` | A ratchet on bare English sentences in the package source. New user-facing sentences belong in the message catalog, where they can be rendered per language; the existing ones are being moved in batches. The gate evaluates string expressions statically before counting words (string literals, templates, `+` chains, `.join` on literal or constant arrays, array spreads and `const` bindings followed across modules), so the same sentence written nine different ways is counted as one sentence, and it records, per file, how many times each normalized sentence occurs. A new sentence, or one more occurrence of an existing sentence, fails; fewer occurrences are reported so the baseline can be lowered. Developer diagnostics (thrown errors, console output), property names and type positions are not counted. A second check keeps the set of exported English sentence constants from growing: new sentences must be functions, so a language can be passed. |
 | `scripts/run-subagent-model-word-test.mjs` | The subagent model word resolver. `resolveSubagentModelWord({ word, liveCatalog?, local? })` returns the word to send as `agents[].model` and a closed reading, never a sentence: catalog names are sent as written, tier words, provider aliases and `best` are sent in canonical lower case so the target engine expands them with its own tier table, a provider model id is replaced by its catalog name, a case-insensitive match is accepted only when it is unique, and anything else is passed through for the engine to judge. When the host asserts that it started the engine itself and passes that engine's active tier group (`local`), and that engine's catalog is also known, the reading also previews where a tier word lands, or why it cannot land (inactive tiers, nothing bound down the chain, a binding missing from the catalog); an unreadable tier table or a missing catalog means no verdict, never an inactive one, and the preview never changes the request. `subagentModelWordProblemLine` renders the one sentence for those cases through the message catalog. The gate compares the preview with the installed engine's own tier expansion on two thousand generated cases, proves the comparison catches five known wrong orders, and checks canonical spelling, absence, hostile inputs, frozen readings and the sentences in both languages. |
-| `scripts/run-retry-status-notice-test.mjs` | The retry-line sentence. `retryStatusNotice(status, ctx?)` turns a retry status (or `null` when the row should be cleared, which gives no sentence) into one sentence, a machine-readable way out (`switch_model`, `wait`, `fix_request`, or none) and the phase of the line. The gate pins the full truth table (arm × terminal × provider-condition word × HTTP status × attempt counts) against an independent oracle and a set of literal sentences; the first-token wait reads as a wait and never as an error; absent or invalid positions are read as absent (no zeros, no invented reasons, no quota wording); words outside the engine's closed set read like `unknown` and are never echoed; the way out is given only where it applies; unknown arms read as the default arm; per-key throwing proxies, a single read per key, a frozen result and purity; the sentence table's keys equal the engine's `PROVIDER_CONDITIONS` and no second table exists; sentence hygiene; and an end-to-end run through the real wire projection. |
+| `scripts/run-retry-status-notice-test.mjs` | The retry-line sentence. `retryStatusNotice(status, ctx?)` turns a retry status (or `null` when the row should be cleared, which gives no sentence) into one sentence, a machine-readable way out (`switch_model`, `wait`, `fix_request`, or none) and the phase of the line. The gate pins the full truth table (arm × terminal × provider-condition word × HTTP status × attempt counts) against an independent oracle and a set of literal sentences; the first-token wait reads as a wait and never as an error; absent or invalid positions are read as absent (no zeros, no invented reasons, no quota wording); words outside the engine's closed set read like `unknown` and are never echoed; the way out is given only where it applies; unknown arms read as the default arm; per-key throwing proxies, a single read per key, a frozen result and purity; the sentence table's keys equal the engine's `PROVIDER_CONDITIONS` and no second table exists; sentence hygiene; and an end-to-end run through the real wire projection. `retryStatusParts(status, ctx?)` returns the same facts as separate segments instead of one string: the model name, the condition phrase, the HTTP status, the attempt number and its cap, the wait and limit of a first-token wait, the engine's own words, the retry-at time as a wall-clock millisecond value for the host to count down, a one-sentence note on what the condition was judged from, and the way out. A segment is present only when the engine gave it, and nothing is ever filled in with zero. The one-line sentence is assembled from the same segments, and the gate checks that the line equals an independently written join of the segments across thousands of combinations. |
 | `scripts/run-event-leg-compose-test.mjs` | Event leg composer: one table decides, for each of the three leg kinds (live stream, durable attach, continuation opened by the question bridge), which stages a frame passes between the SDK event stream and the projection, and the composer only takes host ports. The frame-level stages (run and session adoption, sub-agent owner stamping, in-stream question / elicitation / tool approval frames) have pairwise disjoint frame-type domains (recomputed from the source) and run once on every stream, including each continuation the bridge opens. The run a frame belongs to is recorded per frame object and taken as a pair from one source: on a live stream the terminal frame's run wins over the header and the first frame (an idempotent replay found at row creation), a refused submission is reported with the candidate run id it never created and leaves its frame unbound, frames before a run id is known and frames a host copies without the inheritance helper are unbound, and nothing ever falls back to the most recent binding. A type-checker scan traces every place in the bridge modules where a frame flows (yield, return, frame arrays, frame tables, frame parameters) back to its source and accepts only an untouched reference or the inheritance helper; nine planted shapes (literal copies, Object.assign, structuredClone, helpers with inferred return types, concrete frame subtypes, a frame escaping through a local object, a file in a subdirectory) prove it bites. Driven through the real question bridge: a continuation's frames are stamped and swallowed stage by stage, the approval countdown is shown on the live stream and fully dropped on replayed streams, a hand-off moves frame bindings, owner stamps and the bound callback to the new run, rewritten, collateral and synthesized frames answer the run they came from, the terminal port fires exactly once per released terminal frame and never for the parked one, the bridge dependencies are read from the host object as-is (lane accessors not executed, a throwing accessor does not break the stream, private-field getters keep their receiver), and nothing is persisted once the host says the leg is stale or its staleness check cannot be read. Missing question / elicitation / tool-approval ports fail at assembly with a named host_port_missing error; non-HITL frames pass through unchanged and in order. |
-| `scripts/run-upstream-type-shadow-test.mjs` | Same-name types across this package and its runtime upstreams. Every type name exported from the package root that is also exported as a type by a runtime upstream (the packages listed under `dependencies` and `peerDependencies`, including the typed subpath entries of their `exports` maps, with wildcard entries expanded from the files on disk) must resolve to that upstream's own symbol, re-exported. A local declaration under the same name fails the guard even when its shape matches today, because two declarations under one name drift apart on the next upstream change, and so does a re-export of a different upstream symbol under that name; readings this package mints carry a different name (suffix `Reading`). Module augmentations of a runtime upstream from the package sources fail as well. The only registered exception is an upstream that lags behind the package that owns the vocabulary: the row names the owner, which must be a declared dependency other than this package and whose same-name type must be declared inside that package, and the guard re-proves on every run that this package's type equals the owner's while the upstream's type is a strict subset of it. Rows cover one name and one upstream entry each, fail once they are past their due-by version (at most two minor lines ahead), once the premise no longer holds, or once the name no longer collides. Two types are the same only if each is assignable to the other and their structural fingerprints match, so one extra optional member, a readonly modifier or a wider return shape counts as a difference. The ruler proves itself on an in-memory corpus of package and upstream modules and on two planted declarations checked against the real upstream name set. |
+| `scripts/run-upstream-type-shadow-test.mjs` | Same-name types across this package and its runtime upstreams. Every type name exported from the package root that is also exported as a type by a runtime upstream (the packages listed under `dependencies` and `peerDependencies`, including the typed subpath entries of their `exports` maps, with wildcard entries expanded from the files on disk) must resolve to that upstream's own symbol, re-exported. A local declaration under the same name fails the guard even when its shape matches today, because two declarations under one name drift apart on the next upstream change, and so does a re-export of a different upstream symbol under that name; readings this package mints carry a different name (suffix `Reading`). Module augmentations of a runtime upstream from the package sources fail as well. The only registered exception is an upstream that lags behind the package that owns the vocabulary: the row names the owner, which must be a declared dependency other than this package and whose same-name type must be declared inside that package, and the guard re-proves on every run that this package's type equals the owner's while the upstream's type is a strict subset of it. Rows cover one name and one upstream entry each, fail once they are past their due-by version (at most two minor lines ahead), once the premise no longer holds, or once the name no longer collides. Two types are the same only if each is assignable to the other and their structural fingerprints match, so one extra optional member, a readonly modifier or a wider return shape counts as a difference. A row may also be held for the declared support floor: while the installed upstream already matches the package's type but the support floor still lies below the upstream version where that happened, the row stands; it fails as soon as the floor reaches that version, or when the floor cannot be read. The ruler proves itself on an in-memory corpus of package and upstream modules and on two planted declarations checked against the real upstream name set. |
+| `scripts/run-session-id-canonical-test.mjs` | The session-id canonical reader (`canonicalSessionIdOf` returns the canonical lower-case id or null; `sessionIdRefusalOf` says why not: not a string, wrong shape, bad version nibble, bad variant nibble). The rule is the server's own session-id rule, copied once because the server package is neither a dependency nor exporting its judge; the suite pins the copy against the installed server fixture's source text and checks a superset relation on a generated pool (everything the server accepts is returned unchanged, everything returned is accepted by the server, the only deviation being that upper-case hex is folded to lower case). It also checks a table of shapes, hostile inputs that must neither throw nor be touched, and that exactly one regular expression carries the rule. |
+| `scripts/run-print-model-status-test.mjs` | The print-lane model-service status frames. `retryStatusFromArm` reads the internal retry arm (the same reader the interactive lane uses; sub-agent frames and recovered are told apart), and `createPrintModelStatusTracker` keeps the per-turn status ledger for the `-p` stream-json lane: a retryable failure with both counters becomes a `system/api_retry` frame in the reference client's shape (its keys keep their meaning, every extra key is `_sema_`-prefixed, the category word follows the reference client's table), waiting for the first token, counter-less reconnects, give-ups and replays of the same back-off become `system/_sema_model_status` frames, and an error result carries `_sema_last_model_status`. The sentence comes from the retry-line sentence source. The suite drives the real stream projector end to end, checks the reader against the interactive lane, the replay de-duplication, the category table, the key set, the turn lifecycle (including that a synthetic failure row does not clear the ledger), hostile input, and that the new module holds no module-level state. |
+| `scripts/run-subagent-started-projection-test.mjs` | The delegated sub-agent start frame. The stream projector turns the engine's per-cycle `task_started` frame into a neutral internal arm and the adapter turns that into a `subagent_started` chrome event (no transcript row, no adapter state; the event hangs on the delegating card's lane when the frame names the card, otherwise on the main lane), and `publishEngineAgentPanelStart` feeds the start time into the panel reducer as evidence: it only fills a missing `startedAt` and never creates a row or changes how identities merge. The suite checks the projection key by key (absent keys stay absent, bad shapes drop only the key), the end of the dropped-frame trace line for this frame type (a real unknown frame still gets one), the chrome event end to end through the real stream projector and through the real SDK's SSE parser, the reducer scenarios, and, over 600 random frame sequences, that adding start frames changes no row delta except the `startedAt` field. |
+| `scripts/run-settings-hooks-salvage-test.mjs` | The settings-file hooks salvage reader (`salvageSettingsHooksEntries`): one reading of whether a settings document's hooks load as written, load with some pieces dropped, or are rejected as a whole, with the upstream settings-contract classifier as the only judge. A partial load never hands back the original object: the dropped pieces are removed from the document as written and every kept piece is the original object, so fields outside the contract (such as exec-form arguments) still reach the checks that guard them; the result always classifies as loaded. The suite compares the reader with the upstream classifier over a corpus transcribed from the terminal's own salvage tests, checks the pruning rules and the one-sentence-per-file notice (no values echoed, untrusted names made visible and credential-like names masked), and checks that the hooks plan reads every settings layer through it: a rejected layer contributes no hooks, a rejected managed layer closes the whole hooks leg, and row indices still point into the document the host supplied. |
+| `scripts/run-host-turn-settings-test.mjs` | The single assembly entry for host-supplied request settings (`hostTurnSettings`): hooks, the resolved permission snapshot, the web-search setup and the output-style name reach the request body only through its product, which always carries one row per key (supplied, explicitly absent, or not wired at all, the last one with a session notice and a self-check that names it), a turn verdict, and notices. The request builder takes the product and refuses these keys when they are written straight into the input settings. The suite checks that the request bodies match the previous release byte for byte on inputs whose projection did not change, the missing-source and explicit-none rows, the refusal when managed hooks cannot be loaded (thrown only on a lane that carries the refused key behind an open live gate: the one-shot side-channel lane never carries hooks, so it is sent as before), the final check of the hooks plan against the settings-contract classifier before sending, partial layers, and, against the pinned server fixture, that a configuration the classifier accepts but the server's own limits refuse still fails through the generic invalid-request path. |
+| `scripts/run-seat-devices-test.mjs` | The seat's device face. Whether the device face is usable, and what it can do, is read in exactly one way: `devicesFaceOf` takes whether the host wired the device verbs (the seat capability `devices`, or on a direct host whether it holds the SDK slice), the engine's device-lane reading when there is one, and the outcome of the last device list, and answers one of unavailable (with a closed reason), probe pending, probe failed, verbs off, or manage. On the seat lane there is no capability reading, so the first device list is the lane probe: until it answers, only that one read-only request is allowed and every write stays closed (`devicesWritesAllowed`). An unarmed verb and a host that reports `false` are the same face with the same sentence. The five seat verbs return the package emitters' outcome shapes, and the outbound gates check every arm as a closed shape (a missing required key, a bad value, an extra key or an unknown arm is refused, naming the path); the one-time enrollment token passes only at the top of the successful enrollment-token outcome. The guard also compares the direct-lane answers with the browser client's previous judgement cell by cell, walks a 140-cell seat matrix against an independent specification, and replays the cold-start sequence. |
+| `scripts/run-seat-engine-target-test.mjs` | The seat's engine target (local or Cloud). The engine target has one state, owned by the host and read on the seat through the store `engineTargetStateStore`; six optional verbs switch the mode, run the registry device sign-in, connect or configure a Cloud engine, and clear it. Credentials never cross the seat: the state and every verb outcome are closed shapes at every level (an extra key is refused, naming its path), credential-shaped key names are refused as a second line, and the one inbound credential is the pasted engine token on the configure request, which the inbound judge checks for shape only and never echoes. Failures carry words, not sentences: `engineTargetRefusalText` and `cloudLoginFailureText` render them, and `cloudLoginFailureOf` reads a sign-in failure from the error's own data fields only (the server's error description is never shown; a client-side deadline counts as an expired code). "The engine changed" has one signal, the state's `generation`: `watchEngineTarget` reports each push with a `switched` flag and an `isCurrent(generation)` check, so a refetch started before a switch is dropped when it returns. `seatMcpReadingOf` turns the package's MCP reading into the seat view (an unreadable reading is never shown as an empty list). The guard covers all of this, including the minimal counterexamples for leaked credential keys and the three restart sequences. |
+| `scripts/run-onboard-judgement-test.mjs` | Adding a model, judged in one place. The decisions for onboarding a model live only in this package: the probe verdict (`probeModelEndpoint` — the verdict request is a streamed one-token request and `ok` is earned only by the model's first token within the deadline; a provider turning requests away is `busy`, headers without a token is `no_token`, a 2xx reply that is not a model answer is `shape`; the model list only feeds a similar-id hint; the key goes only to its own base URL and redirects are never followed), whether to write (`onboardWriteVerdict` — refused when the endpoint proved the setup wrong, and written as unverified only on explicit request when no verdict could be had), the write plan (`onboardWritePlan` — entry name, the entry's own key slot, update in place or create, the default model, and the other catalog entries' problems, with the engine's route-pairing rule through the core value ports), and the two naming algorithms. Writing files, keys, locks and restarts stay with the host. The seat gets three optional verbs under one capability `onboard`, with closed outcome shapes; the API key travels only inbound, at the top of the request, and is never echoed. The guard drives the probe with a fake fetch over both lanes, checks key redaction in five echo forms, the write rule, the write plan cell by cell, the naming algorithms against a frozen reference, and the seat shapes. |
+| `scripts/run-user-attachment-seat-test.mjs` | The request-level attachment seat. Its capability reader has four states (not observed, unreadable, not reported, seat) and one predicate decides whether the seat may be used: it must accept entries on the objective, carry text content and accept client text. The suite checks the reader against malformed shapes and against the installed server's own seat description, then checks the encoder: with no usable seat the encoding is byte-for-byte the same as the previous release's encoder on a fixed corpus; with one, whole @-mentioned text files, complete notebooks and fully listed directories move into seat entries named by their full path (partial selections, restored material and other kinds stay in the text envelope), within the seat's per-entry and count limits, never on an otherwise empty message, and inside the request body budget. Entries the encoder emits are run through the engine's own screening in process, the engine's whole-window refusal is read back through the refusal reader, and the engine's per-attachment outcomes are projected back with ordinals matching the encoding. |
+| `scripts/run-user-invoked-skills-slot-test.mjs` | The `userInvokedSkills` request slot. Naming skills widens what a run may do without a card, so the value is only given through `userInvokedSkillsDeclaration`, which returns the key when the capability reading is present and otherwise returns nothing plus the reading that withheld it. The suite checks that truth table, the lanes that seat the key (interactive and print, behind the live gate; never the side channel), that an empty list is not sent and a malformed one is refused at assembly, the path from a landed capability report to the request body, and the refusal reader and sentence for an unknown skill name, read from an error the engine itself raises in process. |
+| `scripts/run-mcp-wire-caps-test.mjs` | The local MCP config projection. A server name goes on the wire byte for byte: a name with leading or trailing whitespace, an empty name, or a name that would collide with another entry once the whitespace is stripped is left out (not silently trimmed, which let ` blocked ` reach the wire as `blocked` and slip past a disabled-server list) and is reported by `projectMcpConfigs`, whose `dropped` rows carry a sentence for the user. `elicitation` is judged by one reader for both the local and the cloud-configuration leg (`mcpElicitationVerdict`): a boolean passes through as written (`true` and `false` alike), absence adds no key, and any other value, `null` included, leaves that one server out with a sentence naming the server and the key. A `toolAxes` declaration is judged by the engine's own reader through the core value port (`mcpToolAxesVerdict`): a bad shape, `null` included, leaves that one server out with a sentence naming the server, the key and the fix, the other servers are unaffected, and with no judge installed the declaration passes through unchanged. The suite feeds the products to the live server fixture's request-entry check: the 400 for a bad `toolAxes` is gone, and a non-boolean `elicitation` sent as written would still be refused by the server, naming the key, which is why neither leg sends it. Since 0.100.0 the name, transport and identity-header readers are the settings contract's own judges, copied at build time and read by both the local and the cloud-configuration leg, so the two legs answer every input the same way. A name that is empty, longer than 128 characters or has leading or trailing whitespace leaves its entry out with a sentence built on a bad-name summary that never echoes the whole name; a transport that cannot be used at all (no command, an address that is not an http(s) URL) leaves the entry out and names the key path; and a transport whose optional parts cannot be used (an argument list with a non-string member, an environment or header record that is not plain string data, a member literally named `__proto__`, a non-string principal header) keeps the entry on the wire as the judge's independent copy without those parts and reports what was left out — key paths and member names only, never a value. Setting both identity headers leaves the entry out with a sentence (it used to be skipped silently). The suite feeds these products to the live server fixture's request-entry check on both generations: every product is accepted, while the same declarations sent as written are refused. Nothing is capped at 32 servers any more: forty servers and a skill body above one mebibyte go on the wire as written, servers from 7.132.0 refuse the whole list or body with a field-invalid answer that names the limit, and older servers keep the first 32 on submission as before. A server start timeout passes through as written and a null server list is never produced. |
+| `scripts/run-agents-wire-caps-test.mjs` | The custom agent projection reads each tool list once and sends the description, the system prompt and skill content as written. The three hand-copied server caps are gone from the root entry (and the truncation and skip logic with them): an over-long field now reaches the server, which answers with a 400 that names the field (`agent.whenToUse`, `agent.systemPrompt`, `agent.skills`). A tool list (`tools` or `disallowedTools`) that is written but is not an array, for example an unsplit "Read, Grep" string, is refused instead of silently dropped: the projection throws a `TypeError` that names the agent and the key and carries the assembly-refusal identity, so the readiness gate keeps the turn from being submitted. |
+| `scripts/run-catalog-routing-test.mjs` | A gate for the message-catalog migration. Sentences that move into the catalog must read byte for byte the same in English as before the move, and must exist in Chinese. Each corpus entry (one sentence producer called with one set of inputs) is compared in English, with and without an explicit language argument, against digests taken from the pre-migration build; a sentence the changelog names as reworded is pinned to its new digest instead of being waived. In Chinese every entry must contain Chinese characters, differ from the English text, and contain no run of three English words outside registered protocol terms, quoted foreign values and code spans; no placeholder may be left over and the message ledger must stay empty after the whole corpus has run. A syntax-tree check requires every function that takes a language argument to pass it on to every other language-taking function it calls, and every exported language-taking function must be covered by a corpus (or be listed as pinned by another gate or as a not-yet-migrated placeholder, which the gate verifies). The gate also proves its own detectors on planted cases. Read-only modes print the digests of a build and compare two builds entry by entry. |
+| `scripts/run-compaction-started-projection-test.mjs` | The `compaction_started` frame (a compaction pass announced at its commit point, before the summary call) projected to an internal arm and then to the additive `compaction_started` chrome event — `pass` and `trigger` required (open-set trigger), `startedAt` only when it is a finite number, malformed frames dropped, zero transcript rows, the lane told honestly; checked against the server package's own frame builders (the start frame and the outcome frame carry the same `pass`). |
+| `scripts/run-compaction-status-tracker-test.mjs` | The single reading of "is a compaction running": `createCompactionStatusTracker()` pairs `compaction_started` with the frame that ends the same `pass` (the landed boundary, a failed or mooted outcome, or the end of the stream), never lets a sub-agent's pass light the main lane, ignores outcomes that carry no `pass`, and gives the interactive lane `{ compacting, since? }`; `compactionStatusFrameOf` turns each change into a `system/status` frame in the reference client's shape for the print lane (no heartbeat, no output-size field). The suite covers interleaved passes, the sub-agent counter-example, end-of-stream closing, older servers, hostile input, both lanes fed the same sequence, and the server package's own frame builders. An opening frame whose `trigger` is not a non-empty string is not recognised on any of the three ways the tracker can be fed (the same rule the projection applies): it lights nothing, is not recorded, produces no status frame, and a good frame for the same `pass` still opens. |
+| `scripts/run-decision-in-flight-ledger-test.mjs` | One ledger answers whether a human decision has been handed to the engine and not yet settled (or settled after a read started), for tool approvals, plan reviews, form prompts from MCP servers and whole runs alike, through one public reading `decisionInFlightOf(identity, readMark?)` and one read mark `decisionReadMark()`. Every writer lives in the package: the plan-review delivery path and its hand-over registration, the durable decide operation, the host's card hold when it sends, the live card while its answer is being sent, and the decision sender. The plan-review entries keep their rules byte for byte (the old and new readings answer every checkpoint of the shared vector table identically, also with the other partitions filled to capacity); a read that started before a decision settled reports a stale snapshot, one that started after does not, and a decision proven not to have taken effect never makes a snapshot stale; positive evidence that the run left the gate settles what is open while the duplicate-delivery latch still holds. Three real writers on one run (a durable decide, a host-held card that is sending, a live card that is answering) are each visible at run level and at their own identity, and the run reads clear only after all three settle. The capacity bound never evicts the package's own in-flight entries; malformed identities and read marks never throw. A refused plan-review reopen that used to carry no reason now names one cause from a closed, frozen set of five (`PLAN_REVIEW_REOPEN_REFUSALS`). |
+| `scripts/run-hitl-decision-send-test.mjs` | One sender for a human decision across three answer endpoints: the live tool approval, the durable in-stream ask and a form prompt from an MCP server (`sendHitlDecision`). The body is built once (verdict, edited input, the deny note behind its capability and length gates, the attribution key chosen by capability, and for the durable ask an idempotency key that is the same for the same ask and verdict and differs across asks); the same body goes through the attribution resend ladder (a retired key is renamed once, a direct-door refusal drops the key once, at most three sends, all before the decision is judged) and a bounded transport retry (the first send plus three, within eight seconds, only for failures without a server verdict and for 5xx other than 501); a durable ask that is mid-park is re-sent at most twice to learn where it landed; only the ask-ledger-disabled code falls back, once, to the live endpoint when an approval id is known; a live 202 parking answer is never re-sent. The outcome is always one of five frozen arms with closed cause sets, and a second send to the same gate while the first is in flight sends nothing. Each case is checked on timing (the outcome is returned after the ledger entry settles; an abort during backoff stops further sends), concurrency (same-gate latch across families, independent gates, a conflicting first decision is reported as such and does not mark the gate as still pending) and every failure class, including malformed form answers that send nothing; the package's two existing decision paths and the sender share one attribution ladder. |
+| `scripts/run-mcp-tool-toggles-test.mjs` | Per-tool switches for connected MCP servers. A switched-off tool is keyed as `{ server, tool }`, where `tool` is the engine's full tool name exactly as the session's MCP panel lists it, so a key is built straight from a panel row and nobody parses or rebuilds tool names. `checkMcpToolToggleKey` is the only shape check for a key, and `mcpToolTogglesToExcludeTools(keys, mountedServers)` is the only mapping from keys to the request's `excludeTools`: only servers mounted this turn produce entries, a host that cannot tell what is mounted passes `'unknown'` so every switched-off tool is excluded, each key is read once and the value that was checked is the value that is used, and no engine port is needed. The seat gains `listMcpToolToggles` / `setMcpToolToggle` behind the `mcpToolToggles` capability; the host keeps the switches and adds the mapped names to `excludeTools` on every turn, and the renderer never carries an exclusion list on `sendMessage`. |
+| `scripts/run-host-turn-settings-intake-test.mjs` | The host-settings assembly entry takes raw material and judges it itself. `memory` and `fileHistory` are host-supplied keys like the four settings sub-keys: the assembly input no longer has them (a hand-filled one is refused by name), the entry reads the `--memory` flag word and the settings `memory` section through `memoryOffDeclared`, and turns the file-checkpoint switch plus the engine's capability reading into a row through `fileHistoryTightenVerdict` (an undecidable reading with the switch off means the turn is not sent, with the same sentence as `fileHistoryUnjudgedRefusal`). The web search source is the raw settings section plus the environment, judged by `resolveWebSearchVerdict` inside the entry (a malformed configuration refuses the turn with the existing sentence). `refilterHostSettings(request, fresh, lane)` is the one subtract-only recheck between assembly and dispatch: hooks are intersected with the fresh product, nothing is added, other host keys are left alone, and a fresh refusal stops the send. Every host-supplied key on the request the assembly builds comes from a product row that carries it, and the assembly reads the product once from its own data (the permission snapshot first, turned into its request form at that read; every other value copied once in its serialized form) (an inherited or getter-backed top-level value is refused, never used; for a getter nested inside a value, what it returned when the copy was taken is what is checked and sent) and uses only that read for the permission snapshot, the carried values and the refusal; the live-defaults layer can still put an output style into `settings.outputStyle`, which is a later step. |
+| `scripts/run-terminal-owner-projection-test.mjs` | Terminal ownership on the busy-reject path. When a submission is refused because another run holds the session, the synthetic terminal row and the result frame now carry the same machine-readable owner: `request-not-sent` (with the host's own request id echoed back and the holding run's id), and a stream the host declares as following another run marks every terminal it produces as `followed-run` (with that run's id). Hosts can therefore tell "my message was not sent" from "the run I am watching finished" without guessing from order. Absence keeps the previous meaning (this terminal belongs to the submission's own run). The reader accepts only the closed set, reads own data properties only and never throws. |
+| `scripts/run-selfheal-exit-slot-test.mjs` | Exit slots for a held session. Every sentence in the busy-session family that tells the user how to end the run holding the session now reads one host slot (`endRun`), and the parked-for-wake sentence reads a second one (`wakeSession`, `null` to omit), so a host with its own command no longer has to override whole rows to remove the engine's HTTP path; without a slot the sentences name the engine endpoint as before. A refused steer now carries the engine's error code, and when that code says the run is no longer running the package reads the run once more and, if it has finished, says so in a sentence that agrees with the earlier busy reading instead of quoting the refusal verbatim. No retry and no cancel are ever issued on that path. |
+| `scripts/run-settings-credential-key-paths-test.mjs` | The one judge of which keys in a settings document hold credentials. It returns key paths only, never values, as findings that carry an optional `numericByNameOnly` flag for count-style names whose value is a plain integer (the flag is present only when true, so a restrictive file-mode decision can still count them while a user-facing notice leaves them out). Key names are judged on two deliberately wide tables, one for environment-style names and one for structured-config names; values are judged only through the display layer's known-secret shapes, URL userinfo and credential-bearing query strings, and an all-string array is read as a command line (`--password <value>`, `--token=<value>`, `NAME=<value>`, `Authorization: …`). Empty strings, null and false do not count. The golden cases, the residual boundaries (a single-letter short option is not judged, credentials embedded in command strings match only by known shape, an unprefixed random value under an ordinary name does not match) and the never-throws / never-leaks behaviour on cycles, getters, throwing proxies and very deep nesting are pinned, together with a differential against the settings package's own secret-token check and the reference client's name table when they are installed |
+| `scripts/run-tools-flag-names-test.mjs` | The parser for the `!` exclusion syntax of a `--tools` list. Each value is split on commas and spaces outside parentheses, then every word is classified: a plain word is kept, a `!`-prefixed word is excluded by name, a rule written in parentheses (`!Bash(git *)`) excludes the whole tool and is reported as scoped, a `!` separated from its name by a space is read as attached and reported as detached, and a lone `!` is reported as bare. Kept and excluded words may be mixed freely — there is no rule that a list is all-negative or all-positive — and names are returned as written, never lower-cased or mapped from older names. The guard pins golden cases, hostile input, and a differential run of the reference implementation's own two functions, extracted from its decoded source by coordinates and run in a sandbox, against three thousand generated lists; it also pins that the parser mirrors the tool-list one and not the stricter debug-category filter that looks similar |
+| `scripts/run-injection-lane-notice-test.mjs` | A single pure function that says, in one sentence, when the local MCP servers and hooks sent with a task will actually run somewhere else. It reads the engine's own execution-lane report (the same reading the doctor line uses): only when that report says tools run on a different machine than the engine, and the lane has hands at all, does it produce a sentence, naming the lane and whichever of three injected things are present — local stdio MCP servers, MCP servers that use credentials from this machine, and hooks — in a fixed order regardless of input order. A local lane, a deployment with no execution lane, a missing, unreadable or malformed report, or nothing injected all yield no sentence rather than a guess. The lane word is shown escaped and capped, the sentence exists in both catalog languages, and the source is pinned not to hand-copy lane names |
+| `scripts/run-rule-refusal-sentence-test.mjs` | The single place that words a refused-rule notice. Given the rows of rules that were not sent (a rule plus the reason it was dead: its tool name is not one the engine recognizes, it uses a path starting with the home marker that cannot be anchored, or the engine's rule compiler refuses it, in which case the engine's own code and sentence are used as is), it returns one sentence that names up to five rules each with its own reason and reports how many more there were; the leading count is the number of distinct dead rules, a rule listed twice is named once, and rows whose reason has its own separate notice (empty, duplicate, capability off, and so on) are not part of it. Rule text is shown escaped and capped and is never echoed inside a reason; malformed input yields no sentence rather than a throw, and the sentence exists in both catalog languages |
+| `scripts/run-result-route-projection-test.mjs` | The route a refused request took, carried on the error envelope of a run result frame under a superset key. When a run ends because the model provider rejected the request with a 4xx and the terminal detail names the route (which credential source, optionally the carrier and bag, and the root), the result frame carries those words as a fresh object; extra fields an engine might add are never copied, unknown words pass through as written, and nothing is carried when the detail, the route or its required parts are missing or malformed, so absence never reads as having taken no route. Everything else on the frame — subtype, error flag, error list, error code, terminal reason — is unchanged whether or not the route is present, and only a provider-rejected terminal ever carries it. The key has a single minting site and a row in the superset ledger |
 
 Each suite carries a floor that only moves up — a refactor that stops executing a group of
 assertions is a failure, not a quieter pass. Guards anchor on the **installed artefact's content**