@sema-agent/client-core 0.90.0 → 0.91.0

package/README.md

@@ -35,7 +35,7 @@ Renamed from **`@sema-agent/wire-cc-adapter`** (0.1.x, deprecated — see *Migra
 
 ## Scope
 
-**Version:** 0.90.0
+**Version:** 0.91.0
 
 - **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
@@ -289,11 +289,11 @@ guard still cross-checks the table by name).
 
 | Suite | What it guards |
 |---|---|
-| `scripts/run-client-core-pure-test.mjs` | Consumer-view behaviour of every moved-in module; segment floors (B1–B8 + C) that only move up |
+| `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. |
 | `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 |
 | `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 |
-| `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. |
+| `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 |
 | `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 |
@@ -351,7 +351,7 @@ guard still cross-checks the table by name).
 | `scripts/run-usage-verbatim-channel-test.mjs` | The two complementary usage disciplines (core 3.0.0 metering semantics): the CC `ModelUsage` mirror stays pure (five pinned keys, `totalInputTokens` has no seat), while the sema-owned channel forwards the engine `turn_end.usage` object **verbatim** (six keys, incl. `totalInputTokens`) via `last_turn_usage.engineUsage` / `handle.latestEngineUsage` — honest absence on pre-3.0.0 engines, no fabricated zeros |
 | `scripts/run-plan-review-decide-verify-test.mjs` | `decidePlanReview`'s post-decide honesty ([2315]/[2316], engine RB-471 family): a 2xx from the decide endpoint is **not** a terminal — the wire re-pulls the task status and words the outcome by the real shape (still-locked / legal new gate / genuinely left park / unverified), never claiming success it hasn't earned; when the engine answers that the session's stored resume context cannot be read, the outcome names the unreadable row and says the decision was not applied. Driven against a real fake-engine HTTP server through the shipped dist. The outcome queue item also carries a machine-readable `_sema_planReviewOutcome` (task id, a package-minted dispatch number, decision, effect) so a host can tell which in-flight decision an outcome belongs to without searching the prose; the prose is byte-identical, the number is minted only for a decision the in-flight latch admits, and a caller that passes no metadata gets no key. |
 | `scripts/run-shell-gate-durable-allow-test.mjs` | #110: the durable approval leg for **shell** gates. The tool_end HOLD/REJECT predicate must cover Bash the same way park detection already does (otherwise the park poison frame `Operation aborted` hits the transcript, `endedCalls` swallows the real replayed result, and the user who pressed Yes watches a command that really ran be reported as aborted); a replayed, already-decided park must resume reading the stream instead of being reported as a failed turn; `lastEventId` must track numeric `seq` too. Mutation-proven: each of the three fixes reverted turns the gate red |
-| `scripts/run-hitl-gate-honesty-test.mjs` | [2393] the four HITL disciplines that a passing type-check cannot see. (1) The park predicate and the `tool_end` predicate must cover the **same** set — the park side admits a first-class `kind:'tool_approval'` gate for *any* tool name, and a `tool_end` frame carries no `kind`, so the frame-level judge falls back to the engine's exact abort marker; otherwise the poison frame hits the transcript and `markEnded` swallows the real replayed result (the #110 disease, reopened on kind-only gates). (2) The already-decided identity criterion is **one-shot**: its two inputs are monotonic, so without consumption one successful decide makes every later park failure — including a real `approvals.list` outage — read as "already resolved" until the 24-hop budget runs out and reports a cause that has nothing to do with what happened. (3) A `plan_review` card dismissed without an answer must be re-presentable: the idempotent re-arm short-circuit re-publishes the still-armed card, and a stale armed id (responder gone) re-arms from scratch rather than presenting a card nobody can answer. (4) `HitlSafetyError` is a safety signal — the `remember` fallback arm must re-raise it instead of auto-retrying the decide, while a plain unknown-key 400 still falls back. (5) The polling leg reschedules after an escaping throw and flips `mode()` to `idle` once it consistently fails, so the honesty surface stops reporting a dead feed as live. (6) The live-frame leg carries the fact behind "you are being asked because the auto-mode classifier could not run" all the way to the card port. Transit narrows on SHAPE only — a non-empty cause string is taken verbatim, an open set, because the word table's owner is the engine and re-checking a closed table at the package boundary would drop a legal value the day a new cause word appears, which is exactly the information worth keeping. A malformed carrier degrades to absence rather than half-minting, and absence stays absence: it covers "the classifier answered", "this ask never qualified" and "this deployment has no classifier" at once, so nothing may render it as reassurance. The guard also pins the division of labour that makes the open set safe — the same word that transits is judged again by the public display reader, which narrows to the availability axis, so a word the engine says it never stamps on this fact renders no sentence while still being visible on the card for triage. A later section pins the split this release introduced on the deny close-out frame. Until now every denied tool call was stamped with the same sentence — the one that says *the user* does not want to proceed — including the calls denied automatically on a lane that has no approval surface at all, where nobody was ever asked. The guard drives all three shapes (a person pressed No, a rule settled it, nobody said which) through both close-out arms and the durable park leg, and pins that the third shape is byte-identical to the previous release: an attribution nobody supplied is not evidence for either answer. The rule-settled shape carries the shell's own reason on a second line when there is one and stands alone when there is not, because a blank line where a reason should be reads worse than no line at all. The attribution is read from own data properties only, so neither a polluted prototype nor a getter can make an automatic denial claim a person made it — and the getter case is pinned to never run at all. The transcript classification word is minted only on the two paths where the upstream transcript format really carries one; the three classifier words and the two abort words are left absent, with the abort words pinned against the strings this package actually normalises interruptions to, which are different strings. A final section pins the decide-operation observer: one `start` in the same tick as the first request and exactly one `end` after the last attempt has settled, across success, retried timeouts, exhausted transient failures, semantic refusal, binding mismatch and both kinds of caller abort, with nothing between retries; an observer that throws or rejects — even when logging that fault fails — never changes what is sent or returned, and the stream-level dependency reaches both durable park legs. A package-internal re-delivery of the same decision (plain approve after an older server rejects the session-scope flag, or a re-send without the attribution key) is reported as one operation with a single start and end. An operation handle only groups sends for the same session and bound call — a send for another gate through the same handle is its own operation — and closing a handle while a send is still in flight defers the end until that send settles. When a person picks "allow for this session" on a parked card and the grant is known not to have been stored — the server answers so, which newer servers do for the gates they can recognise from the parked row as needing a person each time, or the server refuses the session-wide grant with one of the refusal codes that are fixed by the row or the deployment and the package falls back to a plain approval — the parked path now says so with the same line the live path uses, and the receipt carries the server's bit for hosts that call that path directly; an answer without the bit, or any other failure — including a conflict that an internal retry can hit after an earlier send already stored the grant — is treated as unknown and says nothing, a plain approval that never asked for the grant says nothing, and a host logger that throws after a successful decision, on either card path or in the bridge's retry step, can no longer turn it into a second send or a failure. From 0.83.0 it also pins approval cards whose arguments are not the tool's real input: a live frame whose arguments were omitted over the size cap or never sent (the card holds at most a path recovered from the question), and a parked approval whose stored input is missing or replaced by the upstream size marker. On every such card an edited approval sends nothing at all — no respond, no decide, no session-grant attempt — ends as edit-refused, raises the same notice once and flags the card so hosts do not offer editing; both production entry points are driven end to end, and the parked path neither mistakes it for an already-decided gate nor reconnects. The explanatory line is pinned word for word in its four forms: a recovered path is the only thing it claims to have recovered, an unrecoverable one says nothing was recovered, neither claims a reconstructed diff, and the parked forms each say which of the two gaps it is, with near-miss shapes of the size marker still rendered as ordinary input. Real input on the stream, the frame or the record keeps edits flowing, and plain approvals and denials are untouched. Since 0.83.2 the live frame's `mandate` word reaches the card only when it is one of the six known words, read back identically by `readApprovalMandate`; it never adds the separate mandated key to the card, but a known word on its own makes `approvalIsMandated` answer true, because the engine only sends the word as the whole reason for that bit. The frame's `origin` is read from its own keys only, so a value inherited through the prototype chain never reaches the card. |
+| `scripts/run-hitl-gate-honesty-test.mjs` | [2393] the four HITL disciplines that a passing type-check cannot see. (1) The park predicate and the `tool_end` predicate must cover the **same** set — the park side admits a first-class `kind:'tool_approval'` gate for *any* tool name, and a `tool_end` frame carries no `kind`, so the frame-level judge falls back to the engine's exact abort marker; otherwise the poison frame hits the transcript and `markEnded` swallows the real replayed result (the #110 disease, reopened on kind-only gates). (2) The already-decided identity criterion is **one-shot**: its two inputs are monotonic, so without consumption one successful decide makes every later park failure — including a real `approvals.list` outage — read as "already resolved" until the 24-hop budget runs out and reports a cause that has nothing to do with what happened. (3) A `plan_review` card dismissed without an answer must be re-presentable: the idempotent re-arm short-circuit re-publishes the still-armed card, and a stale armed id (responder gone) re-arms from scratch rather than presenting a card nobody can answer. (4) `HitlSafetyError` is a safety signal — the `remember` fallback arm must re-raise it instead of auto-retrying the decide, while a plain unknown-key 400 still falls back. (5) The polling leg reschedules after an escaping throw and flips `mode()` to `idle` once it consistently fails, so the honesty surface stops reporting a dead feed as live. (6) The live-frame leg carries the fact behind "you are being asked because the auto-mode classifier could not run" all the way to the card port. Transit narrows on SHAPE only — a non-empty cause string is taken verbatim, an open set, because the word table's owner is the engine and re-checking a closed table at the package boundary would drop a legal value the day a new cause word appears, which is exactly the information worth keeping. A malformed carrier degrades to absence rather than half-minting, and absence stays absence: it covers "the classifier answered", "this ask never qualified" and "this deployment has no classifier" at once, so nothing may render it as reassurance. The guard also pins the division of labour that makes the open set safe — the same word that transits is judged again by the public display reader, which narrows to the availability axis, so a word the engine says it never stamps on this fact renders no sentence while still being visible on the card for triage. A later section pins the split this release introduced on the deny close-out frame. Until now every denied tool call was stamped with the same sentence — the one that says *the user* does not want to proceed — including the calls denied automatically on a lane that has no approval surface at all, where nobody was ever asked. The guard drives all three shapes (a person pressed No, a rule settled it, nobody said which) through both close-out arms and the durable park leg, and pins that the third shape is byte-identical to the previous release: an attribution nobody supplied is not evidence for either answer. The rule-settled shape carries the shell's own reason on a second line when there is one and stands alone when there is not, because a blank line where a reason should be reads worse than no line at all. The attribution is read from own data properties only, so neither a polluted prototype nor a getter can make an automatic denial claim a person made it — and the getter case is pinned to never run at all. The transcript classification word is minted only on the two paths where the upstream transcript format really carries one; the three classifier words and the two abort words are left absent, with the abort words pinned against the strings this package actually normalises interruptions to, which are different strings. A final section pins the decide-operation observer: one `start` in the same tick as the first request and exactly one `end` after the last attempt has settled, across success, retried timeouts, exhausted transient failures, semantic refusal, binding mismatch and both kinds of caller abort, with nothing between retries; an observer that throws or rejects — even when logging that fault fails — never changes what is sent or returned, and the stream-level dependency reaches both durable park legs. A package-internal re-delivery of the same decision (plain approve after an older server rejects the session-scope flag, or a re-send without the attribution key) is reported as one operation with a single start and end. An operation handle only groups sends for the same session and bound call — a send for another gate through the same handle is its own operation — and closing a handle while a send is still in flight defers the end until that send settles. When a person picks "allow for this session" on a parked card and the grant is known not to have been stored — the server answers so, which newer servers do for the gates they can recognise from the parked row as needing a person each time, or the server refuses the session-wide grant with one of the refusal codes that are fixed by the row or the deployment and the package falls back to a plain approval — the parked path now says so with the same line the live path uses, and the receipt carries the server's bit for hosts that call that path directly; an answer without the bit, or any other failure — including a conflict that an internal retry can hit after an earlier send already stored the grant — is treated as unknown and says nothing, a plain approval that never asked for the grant says nothing, and a host logger that throws after a successful decision, on either card path or in the bridge's retry step, can no longer turn it into a second send or a failure. From 0.83.0 it also pins approval cards whose arguments are not the tool's real input: a live frame whose arguments were omitted over the size cap or never sent (the card holds at most a path recovered from the question), and a parked approval whose stored input is missing or replaced by the upstream size marker. On every such card an edited approval sends nothing at all — no respond, no decide, no session-grant attempt — ends as edit-refused, raises the same notice once and flags the card so hosts do not offer editing; both production entry points are driven end to end, and the parked path neither mistakes it for an already-decided gate nor reconnects. The explanatory line is pinned word for word in its four forms: a recovered path is the only thing it claims to have recovered, an unrecoverable one says nothing was recovered, neither claims a reconstructed diff, and the parked forms each say which of the two gaps it is, with near-miss shapes of the size marker still rendered as ordinary input. Real input on the stream, the frame or the record keeps edits flowing, and plain approvals and denials are untouched. Since 0.83.2 the live frame's `mandate` word reaches the card only when it is one of the six known words, read back identically by `readApprovalMandate`; it never adds the separate mandated key to the card, but a known word on its own makes `approvalIsMandated` answer true, because the engine only sends the word as the whole reason for that bit. The frame's `origin` is read from its own keys only, so a value inherited through the prototype chain never reaches the card. The already-resolved verdict a decision leg uses (`isAlreadyResolvedFailure`: the closed failure code first; codes that say the gate is still standing or failed to start never fall through to the wording check; the moved-gate codes by code; then the four-word wording fallback) is exported from the root entry as the same function the package calls, and a malformed or hostile input answers `false` without throwing, since a `true` would let a caller silently re-attach. The same section compares it, input by input, with the copy a client kept of the same combination, and requires every disagreement to come from one of the four extra checks that copy adds. It also replays the documented replacement order (the closed failure code first, answered by this function; then the copy's extra checks; then this function) and requires it to reproduce the copy's answers on every input, while moving the extra checks ahead of the closed code must disagree somewhere. A refusal that says the stored form of a parked gate cannot be continued while the gate is still pending is classified by its code and its `retriable` bit as a gate that is still standing, so failure text that happens to contain words from the wording fallback no longer makes the leg re-attach; both decision legs carry the bit on their failure outcome. |
 | `scripts/run-park-hop-progress-test.mjs` | L-80: the park re-attach loop budgets **stalled** rounds, not parks. A turn where the model keeps hitting gates and every one of them is really decided (a card was answered, the engine really moved on) must never be cut off by the hop budget — the budget counts consecutive rounds that produced no progress, and "the engine revived and immediately parked again on the same coordinates" is not progress. The three non-progress arms (already-resolved, decide-transport-exhausted, and a re-scan that was adopted but led nowhere) share one same-cause limit instead of one arm having a limit and the others having none, and every non-progress re-attach is announced once through the host callback rather than only to the debug log. When the limit is spent the resolver reads the approval queue once more and puts whatever is decidable in front of the user before it gives up; only when there is genuinely nothing to show does it fail soft, and the terminal message then carries the real cause and a real way out instead of a sentence about a budget. On the self-heal side, a reopen verdict that reports `decidedWithoutCard` — the chain settled the gate by rule, so there was no card to present — is progress, not a reopen failure, and the user is not told their message was NOT sent. Negative control: a genuinely empty queue with a run that never moves still fails soft |
 | `scripts/run-notif-fleet-honesty-test.mjs` | [2393] the five notification/fleet disciplines a green type-check cannot see, each proven by reverting the fix. (1) The workflow-side dedup `return` keeps a count and a trace — without it "suppressed by design" and "a real completion swallowed because the runId minting changed" are the same observation. (2) `seq` normalisation has exactly one mint point, so a 0-based or fractional wire `seq` cannot make the watcher lane and the frame lane key the same completion differently (which would feed the model twice). (3) The TTL sweep defers to a probe arm that is still inside its own deadline — an entry recorded as "abandoned" must not be delivered a moment later — while an arm that has outlived its deadline never blocks the sweep, so the headless exit gate keeps its liveness. (4) The reset hook really clears every ledger it claims to (the sticky `prompt` ledger leaked across cases). (5) The fleet ledger counts all three drop paths (malformed / unknown frame type / isolation drop), and the panel projection's settled recycling is anchored on the settle instant and skips still-present rows, so the dedup token is never carried off with the entry (which would re-emit `end`) |
 | `scripts/run-public-surface-test.mjs` | The outward promises: the npm export surface baseline (an **exact set**, both directions — a new export that never entered the baseline is one nobody watched leave, and deleting it later would not be red), the peer floor witness, and this README's claims. The floor witnesses are a list since 0.90.0 (the installed SDK plus the 15.x and 16.0.0 probe trees); every listed SDK is checked for the value-level symbols and the cost field, and a missing 16.0.0 witness is reported as a skipped section rather than as green. |
@@ -361,10 +361,10 @@ guard still cross-checks the table by name).
 | `scripts/run-client-core-typeshape-test.mjs` | Type discipline as a guard rather than a build side effect: the set of enabled strict knobs (one silently switched off is red), `tsc --noEmit`, and export-surface ratchets for inline anonymous shapes (≥3 members), `unknown` leaving the surface, and bare `unknown` returns — zero slack in either direction |
 | `scripts/run-client-core-singleton-test.mjs` | Module-level singletons ⇄ `docs/refactor/p1-scan/singleton-manifest.json`, **both directions**: an unregistered singleton is red (registering it forces someone to answer "what if this got duplicated"), a stale entry is red, and the `dupRisk: high` count only goes down |
 | `scripts/run-catalog-loader-gates-test.mjs` | The model-catalog candidate chain (`loadCatalogWithSources`) and the provider device-code seam: offline ⇒ `bundled` with an honest `online.reason`, a good source ⇒ `online` plus a cache write, a second offline run ⇒ `cacheHit`; the three hostile source shapes (malformed JSON, `schemaVersion: 99`, off-domain `http`) each fall through to the bundled table, and an off-allowlist target is **never dialled** — including a `302` to another host, proven by a real loopback server's hit counter staying at zero; a one-byte edit to `catalog.sha256` drops that source while an unavailable sidecar only warns; and the device-code poller's `pending → ok` / `expired` arms run against a real loopback HTTP server with an injected clock |
-| `scripts/run-abortable-sleep-test.mjs` | The shared `abortableSleep(ms, signal)` leaf (consumed by `workflowClient.ts` and `agentSession/backgroundView.ts`'s poll backoff): normal timeout resolution, immediate wake-up on `abort` mid-wait, `clearTimeout` really firing on that path, and a post-resolve late abort staying a no-op |
+| `scripts/run-abortable-sleep-test.mjs` | The shared `abortableSleep(ms, signal)` leaf (consumed by `workflowClient.ts` and `agentSession/backgroundView.ts`'s poll backoff): normal timeout resolution, immediate wake-up on `abort` mid-wait, `clearTimeout` really firing on that path, and a post-resolve late abort staying a no-op. Since 0.91.0 it is exported from the root entry, and the guard pins that the root export is the very function the package's own loops call rather than a second copy |
 | `scripts/run-durable-card-display-keys-test.mjs` | The durable approval row's two display keys survive the row→card recast in `surfaceFsApprovalAndDecide`: `governanceForced` stamps on strict `true` only (absence is "no evidence", never `false`), the row's rule offers (`ruleOffers`, or the older `ruleSuggestions` key that earlier servers send) pass through the same shape-narrowing reader as the live-frame leg and land on the **read-only** card key `ruleOffersReadOnly` — plus a standing pin that the durable leg never stamps the redeemable `ruleOffers` card position (the `/decide` body has no rule slot; offering a "don't ask again" option there would be an affordance nothing can honour), and a section for the parked twin of the classifier-unavailable fact: the upstream declares that key on the parked action itself, verbatim and under the same name as the synchronous ask, so this leg reads it rather than guessing a carrier name the way the deliberately unprojected keys must. The guard drives both legs with the same cause and asserts the card ends up byte-identical either way — the observable consequence of one reader serving two key paths, and the thing that silently diverges the day someone writes a second copy. Its own reach is printed rather than implied: what is proven is the package-boundary promise "on the row ⇒ on the card", not that today's engine flattens that key onto the pending row. A further section covers the two display facts the recast had been dropping for far longer. One of them the row has carried all along under a DIFFERENT NAME than the live frame uses — the frame puts it at the top level, the row nests it under the risk descriptor — and that difference in name is exactly why it went unnoticed; unlike the keys this leg deliberately refuses to project, its carrier is witnessed in the engine's own artefact rather than guessed. Neither is decoration: the shell's stand-aside arm reads them, so a call that matched a remembered allow rule which could NOT silence it looked like an ordinary ask on the durable path and was auto-approved with no card at all. Both land on the SAME card slot the live leg uses (one shape for the ends), verbatim bytes, present only when non-blank, never folded into an empty string — and the guard pins the discipline in both directions, including that a top-level key the upstream row does not actually have must still not grow this position. The security-class approval bit (`requiresRealApproval`) rides a parked row's card when the row carries it at top level, on strict `true` only, while look-alike nested carriers are ignored; this is pinned with a constructed row, because today's pending list does not carry the bit yet. A second, separate bit (`irreversibleParkGate`) marks a parked card whose row sits on the irreversible-ask gate kind — a gate-kind fact that covers asks the engine flagged for real approval at the first decision plus safety-tightened gates, with a known engine gap for approval demands raised only on a storage recheck — on the park path only, on the exact gate word only, and never in place of the real bit. Since 0.83.2 the recast also carries the row's ask origin (`origin`) exactly as the live-frame leg does — a non-empty string, verbatim, open vocabulary, never invented or defaulted — and the word a mandated question stands on (`mandate`), accepted only when it is one of the six known words and read back through `readApprovalMandate`; a word outside that set, or a malformed value, leaves the card without it. The word never adds the separate mandated key to the card, but a known word on its own makes `approvalIsMandated` answer true, because the engine only sends the word as the whole reason for that bit. An `origin` inherited through the row's prototype chain is not read, and since 0.83.4 neither is an inherited `mandated` on either leg — both legs stamp it from an own strict `true`, the same reading the seat crossing uses; the single judge `approvalIsMandated` reads `mandated` and `ruleOffersAbsence` the same way, so an inherited key no longer makes it answer true. |
 | `scripts/run-session-memory-status-test.mjs` | The session **memory-status** read face (S-53): the two judgements three clients would otherwise each get wrong. First, *same status, different code* — this route's 404 carries two unrelated meanings (`not_found.session` = unknown or non-owned session; `not_found.route` = a pre-7.53 server that has no such route at all), so dispatching on the **status** would report "your deployment lacks this surface" as "your session does not exist". The verdict is anchored on `errorCode`, the two 404s are pinned to **different** verdicts, and — the load-bearing negative control — a 404 carrying **no** code falls to `failed` rather than guessing either way, since a wrong guess in either direction is a false statement a user would act on. 501 is allowed a codeless fallback because both of its arms mean the same thing here, and `capability.*` stays split from `feature.*` because those two share a status while their dispositions are opposite. Second, *absence means something different per key*: `optOutSource` and `lastCaptureAt` are legitimately absent on a **healthy** session (a zero-history session really is `{captureOptedOut:false, committedCount:0, foldedCount:0}` with no degradation at all), so reading absence as "off/none/0" asserts something unprovable. Two combined readers are pinned: capture opt-out is read from **both** its keys (a record-store fault yields `indeterminate`, never `active` — the difference between "your conversation is being remembered" and "nobody knows"), and last-capture is a **three-state** read whose discriminator is the *other* key, because `lastCaptureAt`'s absence alone covers both "ledger unreadable" and "genuinely no contributions" and therefore decides nothing; the two shapes are pinned to different verdicts so a single-key read turns red. The thin wrapper is the only IO: it never throws, drops malformed keys to absence rather than trusting them (an unreadable value must answer "don't know", never render as truth), refuses to spend a request on an empty `sessionId`, and passes `signal` through untouched |
-| `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. |
+| `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. |
@@ -373,13 +373,13 @@ guard still cross-checks the table by name).
 | `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. |
 | `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 |
+| `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. |
 | `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. |
+| `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. |
 | `scripts/run-submit-wiring-manifest-test.mjs` | The non-streaming submit receipt can carry the run's opening wiring manifest (`TaskResult.wiringManifest`, additive on newer servers). `readSubmitWiringManifest` answers one of three: the key is absent on the receipt itself (older server, or a deployment whose engine never produced that frame) — not the same as unreadable; the key is present but cannot be read (not an object, or none of the twelve sections survive); or a manifest view. The view is the same shape the streaming lane's chrome event carries (minus its two envelope keys) and is assembled by the same code path, so both lanes agree byte for byte on the same object. Liveness fields ride through untouched — this reader never mints a liveness verdict — and an operator-shaped receipt with extra governance sections reads to the same view as a tenant-shaped one. A zero-tool roster is a real reading, not an absence. A receipt carrying only the `instructions` section reads as a manifest, an explicit empty selection is kept, and absence or a malformed section leaves the key out. |
 | `scripts/run-rule-offers-reader-test.mjs` | The narrowing reader behind the "don't ask again" options, now a public entry point rather than a card-port-only one. Hosts that render the frame themselves (a browser has no three-way terminal card) previously had to rebuild this reader on their side, and what it carries is a **redemption-safety** judgement, not a convenience: the batch arm is redeemed by **index**, so a reader that compacts the array after dropping a malformed entry makes the k-th option a person clicked and the k-th rule the server writes two different rules. So: a bad entry is dropped **on its own** (one bad option must not make a real one disappear) while every surviving entry keeps its **original wire index** — pinned from both ends, with the bad entries leading and trailing. A batch's *members* are the opposite: any malformed member drops the whole batch, because a conjunctive batch is one "yes" to all of them and a batch missing a member is a different grant; its honest-remainder count is a reading, not decoration, so a non-integer or negative value drops the batch rather than rendering a fabricated zero. An empty array, a non-array, an over-cap array and an all-bad array all read as **absence** rather than an empty list, because an empty list renders as "there is an option lane with nothing in it". The two wire generations are ordered by a rule, not a preference: the newer key wins outright, a newer key that is **present but unreadable** does not fall back to the retired key (borrowing the older material would pass someone else's options off as this request's), and a `null` newer key reads as absence so a relaying layer that serialises "missing" as null cannot delete the whole lane on older engines. The public entry is finally reconciled against **both** card-port legs on the same material, byte for byte, so the exported reader and the one the card sees can never become two. Two upstream vocabularies used to be **hand-copied** here, and both had fallen behind: a match word outside the copied pair dropped an otherwise valid option outright, and a batch carrying a directory-read member — a member kind the copy did not know — dropped the whole batch. Both tables now come from one place upstream and are re-exported verbatim, pinned in both directions: every word in the table must be accepted (a narrower copy reds on the words it never learned) and a word constructed to be outside it must still be refused (a reader widened to "any string" reds too), with the retired-key normalising leg sharing the same narrowing so the fix cannot land on one leg only. A member whose kind is genuinely unknown still drops **the whole batch and only that batch** — never one member, because a conjunctive batch one member short renders "yes to N" as "yes to N−1", and never the card, because the honest single beside it is intact — while a member from before the discriminant existed normalises to the historical kind rather than being refused. The additive per-segment reasons ride through verbatim, drop only the row that is malformed, and stay **absent rather than empty** when nothing survives, since an empty list would read as "confirmed nothing uncovered" while the count remains the only source of truth |
-| `scripts/run-resume-refusal-copy-test.mjs` | The **words** a client says when a resume is refused, minted once here instead of three times. The facts behind them already lived in this package; the sentences did not, so each client wrote its own — and those sentences answer a safety question (was my decision consumed, can this token still be redeemed), which is exactly the kind of answer that must not vary by client. Two closed sets meet here and the guard pins their relationship in both directions, because it is a premise rather than a coincidence: one set answers *can waiting help* (the codes the server mints a wait on), the other answers *what should a person be told*, they **intersect in exactly one code**, and each keeps a member the other must not have — a placement mismatch is never waitable no matter what arrives on the response, since its remedy is a changed argument rather than elapsed time, and a full governance window needs no prose because "you can wait" is the whole message. The overlapping code delegates its wait and its disposition to the existing reading rather than judging again: nine shapes of input drive both entry points and the two readings must agree byte for byte, the absent case included, because two judges always diverge somewhere. The wait is narrowed to the domain the server mints it in, which is **stricter than the shell's own copy was** — a zero now reads as no window rather than as "retry now", and the wake-up it would retry is an at-most-once action with real side effects. The third sentence is chosen by the disposition, never by the engine's prose: rewriting the message to either upstream branch's exact wording, with the window untouched, must leave all three sentences unchanged, while adding a window must change the third one and only the third one. The engine's folded resume refusal (`resume_blocked_by_policy`) gets its own reading — the original code as named, absent or unreadable — and a wording that neither claims nothing was consumed nor predicts whether a retry would pass. From engine `7.104` a folded admission refusal always names a code: when the underlying refusal carried none, the server fills in the generic code for its HTTP status. Those generic codes are not a specific check, so the sentence says the refusal did not name a more specific check (and shows the generic code in brackets) instead of presenting it as the check that refused the resume; the generic-code list is compared word for word, in both directions, with the server dispatcher that mints them, item by item and in order with the same-named list the SDK exports from 16.0.0, and word for word with the table in the server's wire contract. The SDK comparison reads the probe trees named by `SEMA_SDK16_ROOT`, `SEMA_SDK15_ROOT` and `SEMA_SDK15_ROOT_OLD` plus the installed SDK; when the full fixture set is required, having no SDK at 16.0.0 or later among them fails the check. Otherwise a comparison is reported as skipped only when its source is not configured or the server source predates the contract section; a configured source that is missing, unreadable or lacks the contract document fails the check. |
+| `scripts/run-resume-refusal-copy-test.mjs` | The **words** a client says when a resume is refused, minted once here instead of three times. The facts behind them already lived in this package; the sentences did not, so each client wrote its own — and those sentences answer a safety question (was my decision consumed, can this token still be redeemed), which is exactly the kind of answer that must not vary by client. Two closed sets meet here and the guard pins their relationship in both directions, because it is a premise rather than a coincidence: one set answers *can waiting help* (the codes the server mints a wait on), the other answers *what should a person be told*, they **intersect in exactly one code**, and each keeps a member the other must not have — a placement mismatch is never waitable no matter what arrives on the response, since its remedy is a changed argument rather than elapsed time, and a full governance window needs no prose because "you can wait" is the whole message. The overlapping code delegates its wait and its disposition to the existing reading rather than judging again: nine shapes of input drive both entry points and the two readings must agree byte for byte, the absent case included, because two judges always diverge somewhere. The wait is narrowed to the domain the server mints it in, which is **stricter than the shell's own copy was** — a zero now reads as no window rather than as "retry now", and the wake-up it would retry is an at-most-once action with real side effects. The third sentence is chosen by the disposition, never by the engine's prose: rewriting the message to either upstream branch's exact wording, with the window untouched, must leave all three sentences unchanged, while adding a window must change the third one and only the third one. The engine's folded resume refusal (`resume_blocked_by_policy`) gets its own reading — the original code as named, absent or unreadable — and a wording that neither claims nothing was consumed nor predicts whether a retry would pass. From engine `7.104` a folded admission refusal always names a code: when the underlying refusal carried none, the server fills in the generic code for its HTTP status. Those generic codes are not a specific check, so the sentence says the refusal did not name a more specific check (and shows the generic code in brackets) instead of presenting it as the check that refused the resume; the generic-code list is compared word for word, in both directions, with the server dispatcher that mints them, item by item and in order with the same-named list the SDK exports from 16.0.0, and word for word with the table in the server's wire contract. The SDK comparison reads the probe trees named by `SEMA_SDK16_ROOT`, `SEMA_SDK15_ROOT` and `SEMA_SDK15_ROOT_OLD` plus the installed SDK; when the full fixture set is required, having no SDK at 16.0.0 or later among them fails the check. Otherwise a comparison is reported as skipped only when its source is not configured or the server source predates the contract section; a configured source that is missing, unreadable or lacks the contract document fails the check. From server 7.115.0 the continue-only-an-existing-conversation refusals carry their own codes (404, and a 422 field conflict) and the reader takes the code first, falling back to the status-derived code plus the trailing marker only for older servers. A parked task the server can no longer continue in its stored form (409 `checkpoint.unsupported_version` with `retriable: true`) has its own reader and sentence, shaped like the reopen family and reading `retriable` the same single way: the sentence says the work is still waiting without promising that resending will continue it, and its way out depends on the parked kind and run id the caller supplies; the plan-review decision and the decide-failure note read this code the same way. For a plan review refused this way, the closing instruction no longer says the plan is still waiting for a decision; it says deciding again will not take effect and points to the way forward, and it comes from the same tail function as every other outcome. |
 | `scripts/run-resume-retry-later-test.mjs` | The two resume refusals that carry a **wait quantity** — the only members of that refusal family that do, which is the whole reason they form a closed set. Carrying a wait is not the same as being the only ones worth waiting on: a sibling refusal in the same family clears on its own and the engine says so in words, it just cannot put a number on it, so *not recognised here* must never be read as *waiting will not help*. One of the two also has a *terminal* upstream branch that arrives under the same code with the distinguishing detail only in prose, so recognition alone is not permission to say "try again": the disposition is decided by **positive evidence** and pinned from both directions — the quota-window code is evidence in itself, the preflight code counts only when the server really supplied a wait (an upstream fact, not a convention: the terminal branch throws with no detail at all, so a wait value cannot reach the client on that path), and a preflight refusal with no wait reads as *undecidable* (say what is true of both branches — nothing was consumed — and leave redeemability to the engine's own line) rather than being rendered as either a retry or an ending. Every other member means waiting will not help (change a setting, relaunch, the retained session is gone), so the recognition is a **closed set of two codes**: widening it to a family prefix would tell half the users to wait and the other half to keep waiting for something that will never arrive, and the negative controls drive exactly those codes through it, plus a same-named code on a different door (the submission-side quota refusal), the two underscore-form siblings, and a code merely quoted inside a message body. The wait value is narrowed to the same domain the server mints it in (a whole number of seconds, at least one): zero, a negative, a fraction and a non-number all read as **no window given** rather than as zero, because a zero tells the caller to retry immediately and the wake-up it would retry is an at-most-once action with real side effects. Reading is structural rather than `instanceof`, since the client is host-injected and the same class name across two bundles is two classes, and a null-prototype plain object must still be recognised. The failure classifier gains this one disposition without any existing one moving, an unknown code still falls to the honest open-set arm and its wait value is **not** believed, and an end-to-end call proves the disposition and the window reach the host while the call itself is still attempted exactly once. The recognised code set is a **frozen array**, not a type-level readonly set: the latter is a plain mutable collection at runtime and the decision reads the same instance, so one `.add` from any consumer would turn a refusal that waiting cannot fix into one that claims it can — the guard proves it by really trying to mutate the exported value and then checking the verdict did not drift |
 | `scripts/run-model-capability-probe-test.mjs` | Whether a model on the OpenAI-completions lane **thinks**, and whether that thinking can be **turned off** — a question nobody can answer by looking at a model name, and one whose wrong answer costs every later call. The probe is judgement only: the network half arrives as an injected port, so the package mints no URL, reads no credential and never calls `fetch` — pinned by a source-level assertion, because a package that reaches the network once has changed what every host must trust it with. The seven dialect words are a **copy**, reconciled element-wise against the installed engine’s own bytes in both directions, since the words belong upstream and a private table drifts the day a dialect is added; the settings package deliberately declines to restate them, so the table cannot be imported from there and this guard is what stands in for the import. The **order** the dialects are tried in is a public promise rather than an implementation detail — each extra attempt is real money and real latency against someone’s gateway — so the guard pins the exact call sequence a stub records, and reversing it reds on the wasted round trip; the template-parameter spelling leads because an observed gateway keeps thinking, and answers with an empty body, when handed the top-level switch instead. That observation is also why an empty answer is **not** accepted as *thinking is off*: a knob that deletes the reply is not a knob that disabled reasoning, and accepting it would write a spelling into the catalogue that the gateway does not honour. Two dialect words whose request bytes are identical to another’s do not each burn an attempt. The two verdicts that look alike are held apart from both directions: *tried everything, still thinking* requires at least one attempt to have **cleanly answered**, and when every attempt was refused the verdict is *could not tell* instead — and on the unanswerable path the result carries **no** thinking flag at all rather than a fabricated `false`, while the pure write-back returns the very same entry object untouched. A verdict that reasoning cannot be disabled **removes** a previously declared spelling rather than leaving it, since a refuted spelling keeps the engine sending bytes the gateway ignores while the catalogue still renders it as already off. Evidence is lengths, finish positions and status codes only — a planted secret in both the answer and the reasoning channel must appear nowhere in the result, so the record can go into a log or a ticket whole. One cross-package premise is checked by really running the other package’s parser rather than quoting its documentation: everything this probe writes eventually passes through the settings schema on its way into a catalogue, and that field is declared parse-transparent precisely so the vocabulary can live on the consuming side. If it ever narrows, the spelling is stripped **silently** — indistinguishable from the probe never having run — so the guard feeds the probe’s real output through the real parser, checks the compat object comes back key for key, and checks a dialect word this client has never heard of survives too. Two shapes that must be rejected really are rejected, since otherwise the survival checks would hold on a parser that accepts anything, and a bare entry is asserted valid first, because the first run of this section reddened on a space in a fixture’s name — a fixture that cannot pass would disguise the real alarm as already having fired |
 | `scripts/run-decide-receipt-test.mjs` | What a decision verb actually **answered** — and, more importantly, what it did not. A success response on the newest lane is only an acknowledgement that the decision was accepted for delivery: the approval is still pending, and a client that clears the card on it shows either a ghost card that was already approved or a card that vanished while the decision was lost. So the package deliberately has **no** "was it resolved" predicate — nothing in that body can answer it — only the opposite one, whose `false` is likewise not evidence of resolution; resolution is only ever the next running arm on the stream. The guard pins that inversion in the product source too: the success path must no longer clear the latched gate, while the stream-observing path that really clears it must still be there. The body has four shapes with **no** key common to all of them, so every position is read as honestly absent, and the handoff handle — which run to watch from here on — requires **two** facts together, since either one alone would either point the stream at the run it already had or mint an empty handle. The record of what finally happened to an already-decided action is read through the **same** reader as every other gate record rather than a second copy, and its absence means **unknown**, never *it was allowed* — the two can even contradict each other, so the card says nothing at all when it is missing. The three refusals on that lane each get one distinct sentence and a disposition taken from **why** each was refused rather than from severity: one cannot be helped by re-sending at all, one waits on the host, one just drops an option — and none of them carries a countdown, because the server never mints a wait for them. Recognition is a **closed set**: an unrecognised code on the same prefix returns nothing rather than a guess, since that prefix also houses a safety signal whose whole rule is never to retry automatically, and the recovery handle is read as absent when unreadable rather than substituted from a different identifier that no longer appears on that lane **0.68.3 (core 7.18.0):** `gate.disposition.classifier` is read key by key into a named view (requested model, model that answered, and the ladder fallback when one happened); a half-shaped record yields no classifier at all rather than a half view, and absence stays "unknown", never "the seat answered itself" |
@@ -407,7 +407,7 @@ guard still cross-checks the table by name).
 | `scripts/run-dist-orphan-test.mjs` | Every `.js` / `.d.ts` under `dist/` must have a same-named source under `src/`, and every source must have its build output — because the compiler only writes and never deletes, so a module removed from the sources keeps shipping from the previous build (the whole `dist/` directory is on the publish whitelist) while the public-surface gate only looks at what the barrel exports and the hygiene gate only looks at forbidden words. Orphans are named one by one; the pre-publish posture is a clean rebuild, and this gate is the check that the posture was actually followed. |
 | `scripts/run-dist-comments-test.mjs` | **dist ships zero comments.** Since 0.77.2 the build strips comments (`removeComments`); this gate walks every shipped `dist/**/*.js` / `*.d.ts` and counts comment trivia with the TypeScript scanner (string literals containing `//` and generator methods are not comments), failing on the first one (`DIST-COMMENT-FAIL`). Source comments are an internal surface; what still ships is code, string literals and type-level text, which the hygiene gate screens. Negative control: one plain comment appended to `dist/index.js` turns it red. |
 | `scripts/run-task-request-omission-receipt-test.mjs` | Where every key a client hands to the request constructor ends up. The constructor used to answer "not stamped" the same way for four different reasons — value absent, no such row, wrong lane, live gate closed — and a key it had never heard of did not even get that: an unattended run could pass a system prompt, an output schema and a spend cap and receive a body holding the objective and the session id, with nothing anywhere saying what was left out or why. The guard pins the three answers apart. **Seated** keys reach the body verbatim on the unattended lane. Keys the package **knows but did not carry** never throw, never reach the body, and each gets a receipt row with one word from a frozen cause list — every present key is on the body or on the receipt, never both and never neither, checked across both lanes with the live gate open and closed against a key-by-key table written independently of the package's own routing. Keys the package **does not know** are refused loudly and are a separate cell, not a fourth cause: the cause list has no word that could hold them, and the seat reader answers `unknown`, not `none`. The cause list is bitten from both sides (exact, every word producible, nothing produced outside it, the judge table's keys read from source through the TypeScript parser) and no second hand-copied list may exist in `src/`. Upstream claims are read straight off the installed SDK typings: a key seated in this release must be a named request field, a key registered as having no upstream counterpart must not be — the day it appears the guard turns red — and the index signature counts as evidence for nothing. The per-run file-history opt-out word is covered the same way: seated on the two user lanes behind the live gate and absent from the side-channel lane, stamped only for its one legal word, refused loudly for any other value, and read after every existing declaration so that it cannot erase them. The wording shown to end users is pinned alongside the wording for integrators: a second sentence table, keyed by the same frozen cause list and exhaustive over it at compile time (the check sits on a two-column table literal — a lead-in plus a reason clause — from which the full sentences are derived), gives each cause one plain sentence that says the item was not sent and why (the integrator table now carries the same literal-level check — before, a missing sentence failed the build but an extra one did not), with no integrator vocabulary and no claim about whether anything took effect; every sentence differs from the others and from the integrator wording and passes the published-text hygiene list, and an unknown word, a non-string, a throwing object or a prototype-chain name gets one honest generic sentence instead of a throw, an echo or an invented cause. The reason clause on its own is available for clients that already print their own "was not sent" line: the full sentences and the generic fallback stay byte-identical to the previous release, every clause starts lower-case with no closing full stop, an unreadable cause gets the generic clause, and no full sentence may appear as a literal anywhere in the source. When a consuming client's own copy of the cause table is on disk, the guard also checks that every word in it is one of the package's causes, with no word listed twice (a word only the client's table has fails; a word only the package has is reported, not judged), and reports that check as skipped rather than passed when the table is not there. A retired request key the engine now refuses by name is refused at construction on every lane, in every live state and through both doors, with a sentence of its own that names the key, says it never authorized anything and points to the permission mode for anyone who wants the engine to release asks; an undefined or null value is read as absence, the seat reader answers unknown, the wire-side judge names it, and a getter on it runs after every declaration so it cannot erase one. |
-| `scripts/run-session-policy-wire-test.mjs` | The per-session tool-rule face: the capability bit that says whether an engine keeps such rules at all, and the narrow read plus tightening orchestration built on it. The bit is read the same four-state way as its sibling capability readers — an absent key is not reported (this binary predates the position itself, which says nothing about whether the face exists), `true` is present, `false` is a positive absent (this deployment keeps no per-session rules), any non-boolean value is unreadable and drops the cell rather than being folded into "absent", and a capabilities body that is not an object at all (an array included) is unreadable rather than "not reported". Its single-source verdict answers whether to show the tightening entry: only an engine that says yes is `yes`, both a positive no and a binary too old to answer are `no`, and never having observed a capabilities body is `unknown`. Whether to put a request on the wire is deliberately a **different** question with a different answer for that last state, and lives with the orchestration. The read narrows three ways that must not collapse into each other: a record that really is empty (present, version zero — what an engine answers for a session no rules were ever written for), a record that cannot be read, and a call that failed with a typed disposition. A half-bad record — one rule bucket well-formed and another the wrong shape — counts as unreadable in full, because the write verb replaces the whole record: dropping the bad bucket and writing the rest back would empty it, which relaxes the rules while the caller sees a 200. An unreadable version stamp is never filled in with a zero, a bucket that reports an implausible number of entries is unreadable rather than walked or truncated, each array's length and each of its indices are read exactly once, and a throwing accessor is unreadable rather than propagated. Every load-bearing key is read as an own property — the envelope, the version stamp, each of the five buckets and each array index — because a prototype lookup would let a polluted prototype put a bucket into the reading that the wire never carried, and since the write replaces the whole record the union would then write that invented restriction back as a real one; a guard pollutes the object and array prototypes in place and proves all four shapes stay out. Because the write replaces the whole record, adding a restriction means writing "what is already there, plus the new entries": the union only ever adds, de-duplicates verbatim, keeps a bucket that is present but empty (present-and-empty and absent are opposite meanings, and dropping it would relax the rules), mints no bucket neither side had, copies entry bytes as they came (no trimming, sorting or path rewriting — those judgements belong to the engine), and takes its bucket names from the engine's own type surface rather than a hand-copied list, so a new bucket upstream is a compile error instead of a silently dropped one. Which differences count as relaxing is the engine's judgement and is never re-implemented here: a refusal on those grounds is reported verbatim, never swallowed and never retried. The orchestration is guarded on three axes. Timing: when the record moves between the read and the write, it re-reads and re-writes **exactly once** — two reads and two writes, no more — and the second attempt's union carries the other writer's entries, which is the entire point of re-reading; a second collision is reported rather than retried a third time, and an uncontended write makes exactly one round trip. Concurrency: an explicit barrier holds both orchestrations first reads at the same version before either may write, and the criterion is how many times the store actually rejected a stale version rather than how many writes it saw — the latter is equally true of two serial successes, so it would stop detecting contention the day the interleaving changed. Under real contention the store rejects exactly once, both writers land on strictly different versions, both writers entries survive in the final record, and the round trips are exactly three reads and three writes; the same two orchestrations run serially are asserted to reject zero times in two reads and two writes, which is what proves those numbers are discriminating. On a store where every write loses the race both report a collision having written exactly twice each. Failure classification: a relaxation refusal, a refusal to stamp a version the store cannot establish (the same status code as the relaxation refusal but a different machine code, and folding it into that arm would send the caller off to edit entries that are not the problem), a missing session, a deployment without the face, an ownerless session, a collision code, a bare conflict with no machine code, a rejected body and an unauthorized call each land on their own arm — the collision arm is matched on the machine code verbatim rather than on the status, because two different situations share that status and only one of them is worth retrying. The remaining split is not "which code is this" but "did the engine answer at all": an answered client-side refusal is allowed to say nothing was written, because every such refusal on this endpoint is emitted before the record is touched, while a throw with no answer at all — a dropped connection, a timeout, a response body the transport itself could not decode, a server fault — can only say "unknown", since that throw may well have happened after the record was already saved. Two guards prove that is not theoretical: a write whose receipt cannot be read, and a write that throws after the fixture store has committed, both leave the record changed. Neither is success nor failure: the only honest answer is "unknown", it carries no version, and it is never retried. The direction of the change is likewise never claimed. The engine’s tighten-only rule is an identity gate, not a field gate — for a principal the deployment treats as an operator it does not run at all, so a union that adds a name to an existing allowlist is accepted and really does widen it. This package does not mint a second copy of that rule, so what it reports is the fact it can stand behind: the record now holds what it already had plus the entries sent here. The sentence for a saved write is pinned to contain no claim of tightening, narrowing or restriction, and a guard reproduces the operator case to prove the widening is real while the wording stays honest. Every sentence the module mints is checked pairwise distinct, with the receipt-unreadable one required to keep its "may already be in effect" and the record-unreadable one required to say nothing was written. |
+| `scripts/run-session-policy-wire-test.mjs` | The per-session tool-rule face: the capability bit that says whether an engine keeps such rules at all, and the narrow read plus tightening orchestration built on it. The bit is read the same four-state way as its sibling capability readers — an absent key is not reported (this binary predates the position itself, which says nothing about whether the face exists), `true` is present, `false` is a positive absent (this deployment keeps no per-session rules), any non-boolean value is unreadable and drops the cell rather than being folded into "absent", and a capabilities body that is not an object at all (an array included) is unreadable rather than "not reported". Its single-source verdict answers whether to show the tightening entry: only an engine that says yes is `yes`, both a positive no and a binary too old to answer are `no`, and never having observed a capabilities body is `unknown`. Whether to put a request on the wire is deliberately a **different** question with a different answer for that last state, and lives with the orchestration. The read narrows three ways that must not collapse into each other: a record that really is empty (present, version zero — what an engine answers for a session no rules were ever written for), a record that cannot be read, and a call that failed with a typed disposition. A half-bad record — one rule bucket well-formed and another the wrong shape — counts as unreadable in full, because the write verb replaces the whole record: dropping the bad bucket and writing the rest back would empty it, which relaxes the rules while the caller sees a 200. An unreadable version stamp is never filled in with a zero, a bucket that reports an implausible number of entries is unreadable rather than walked or truncated, each array's length and each of its indices are read exactly once, and a throwing accessor is unreadable rather than propagated. Every load-bearing key is read as an own property — the envelope, the version stamp, each of the five buckets and each array index — because a prototype lookup would let a polluted prototype put a bucket into the reading that the wire never carried, and since the write replaces the whole record the union would then write that invented restriction back as a real one; a guard pollutes the object and array prototypes in place and proves all four shapes stay out. Because the write replaces the whole record, adding a restriction means writing "what is already there, plus the new entries": the union only ever adds, de-duplicates verbatim, keeps a bucket that is present but empty (present-and-empty and absent are opposite meanings, and dropping it would relax the rules), mints no bucket neither side had, copies entry bytes as they came (no trimming, sorting or path rewriting — those judgements belong to the engine), and takes its bucket names from the engine's own type surface rather than a hand-copied list, so a new bucket upstream is a compile error instead of a silently dropped one. Which differences count as relaxing is the engine's judgement and is never re-implemented here: a refusal on those grounds is reported verbatim, never swallowed and never retried. The orchestration is guarded on three axes. Timing: when the record moves between the read and the write, it re-reads and re-writes **exactly once** — two reads and two writes, no more — and the second attempt's union carries the other writer's entries, which is the entire point of re-reading; a second collision is reported rather than retried a third time, and an uncontended write makes exactly one round trip. Concurrency: an explicit barrier holds both orchestrations first reads at the same version before either may write, and the criterion is how many times the store actually rejected a stale version rather than how many writes it saw — the latter is equally true of two serial successes, so it would stop detecting contention the day the interleaving changed. Under real contention the store rejects exactly once, both writers land on strictly different versions, both writers entries survive in the final record, and the round trips are exactly three reads and three writes; the same two orchestrations run serially are asserted to reject zero times in two reads and two writes, which is what proves those numbers are discriminating. On a store where every write loses the race both report a collision having written exactly twice each. Failure classification: a relaxation refusal, a refusal to stamp a version the store cannot establish (the same status code as the relaxation refusal but a different machine code, and folding it into that arm would send the caller off to edit entries that are not the problem), a missing session, a deployment without the face, an ownerless session, a collision code, a bare conflict with no machine code, a rejected body and an unauthorized call each land on their own arm — the collision arm is matched on the machine code verbatim rather than on the status, because two different situations share that status and only one of them is worth retrying. The remaining split is not "which code is this" but "did the engine answer at all": an answered client-side refusal is allowed to say nothing was written, because every such refusal on this endpoint is emitted before the record is touched, while a throw with no answer at all — a dropped connection, a timeout, a response body the transport itself could not decode, a server fault — can only say "unknown", since that throw may well have happened after the record was already saved. Two guards prove that is not theoretical: a write whose receipt cannot be read, and a write that throws after the fixture store has committed, both leave the record changed. Neither is success nor failure: the only honest answer is "unknown", it carries no version, and it is never retried. The direction of the change is likewise never claimed. The engine’s tighten-only rule is an identity gate, not a field gate — for a principal the deployment treats as an operator it does not run at all, so a union that adds a name to an existing allowlist is accepted and really does widen it. This package does not mint a second copy of that rule, so what it reports is the fact it can stand behind: the record now holds what it already had plus the entries sent here. The sentence for a saved write is pinned to contain no claim of tightening, narrowing or restriction, and a guard reproduces the operator case to prove the widening is real while the wording stays honest. Every sentence the module mints is checked pairwise distinct, with the receipt-unreadable one required to keep its "may already be in effect" and the record-unreadable one required to say nothing was written. When a read comes back without a usable record, one sentence per failure kind and per unreadable position says so in plain words, never naming internal codes, never saying the record is empty, and falling back to one general sentence for anything it does not recognise. |
 | `scripts/run-persisted-rule-write-test.mjs` | The **single-step tightening write** for persisted permission rules — the dual of the revoke surface, and the half where a hopeful reading is expensive. The two behaviours this entry accepts are **derived** from the three-state vocabulary by subtracting the widening one, never hand-copied: the guard bites in both directions (every word in the derived table is really accepted; every constructed outsider — casing variants, trailing whitespace, the widening word itself — is refused before a single round trip), keeps a word-count canary against the parent table, and pins that the source file contains **exactly one** array literal carrying two or more behaviour words, so a second hand-written table shows up as a boundary failure rather than as drift nobody reads. A standing approval is minted by answering a permission question or by importing settings; this entry is not a third route, and the widening word is unspellable in the type. The outcome is a discriminated union whose two failure arms are **not** interchangeable: ten refusal causes each promise the same single thing — not one byte reached the store — while three separate words say the opposite, that the outcome could not be read at all. The service's own "I cannot tell" (a write that could not be confirmed as standing: store wobble, a redemption leg with no decidable ending, or a write that landed and was revoked concurrently before the read-back) stays in the second group, because announcing "nothing was written" invites a clean retry that is not clean, and announcing success misreports a tightening that may already be gone. Anything the shared failure classifier does not recognise defaults to the same place — this is a non-idempotent verb, so "unclassified" must mean "unknown", never "no write": a 500 can happen after the store commits. A 2xx whose body cannot be read is pinned in the same direction and from both sides: it reads as unknown, and the unknown arm carries **neither** the revision nor the written row, so a consumer cannot even spell the shape that would let "unreadable" pass for "written". `persisted` is guarded against the reading everyone reaches for first: it says *this call wrote*, not *a new rule now exists* — an equivalent rule already in the store still mints a fresh causal point, so the lane honestly reports `persisted`, and the material for judging whether the **logical** rule is new (the approval ledger on the returned row) is handed to the caller rather than folded into the discriminant, since the package does not have the one fact that judgement needs. The returned row goes through the **same single narrower** the listing surface uses — proven by running one row corpus through both legs and asserting the two verdicts agree entry for entry (an adversarial pass that forks the listing leg back into an inline copy reds here immediately), plus a source pin that the predicate is defined once and called from exactly the two legs. That sharing is what keeps a row whose behaviour cell is unreadable **visible in both places** rather than hidden by one of them — and the shared narrower is deliberately followed by a second, *different* question only the write leg can ask: is the row that came back **the rule that was just sent**? A rule's identity is a triple, so a receipt missing its behaviour cell, carrying the sibling state, carrying the widening one, or naming another text or another scope is not evidence that the requested tightening is standing; it reads as unknown with its own word, kept distinct from "unreadable" so the two stay tellable apart, while display and derived cells may vary freely. An adversarial pass found both of the gaps this pins: the receipt check that only looked at whether the row was renderable, and a subtler one — pulling the verb off the injected port and calling it bare drops the receiver, so a host that hands over a real resource object (a class instance whose verbs reach the transport through `this`) would see every write throw and be reported as "could not tell", retry after retry, while the package's three other ports call their verbs as methods and work fine. Both are pinned from the failing side: a shorthand-method facade and a class-instance facade must reach the transport and return a real outcome, with the bare-call throw proven to be a real failure mode first. A 405 is split in two, because only the engine's own bare code is evidence about **the engine**: with it, the path exists and this verb does not, so this worker predates the verb; without it — an absent, empty, or foreign code, which is what a proxy or gateway blocking the method typically returns as HTML or an empty body — what was seen is that the verb was refused, while **who** refused it and **at which hop** is unknown, so it lands in the unreadable-outcome arm with its own word rather than sending someone to upgrade a worker that is fine, hiding an entry that is live, or — the part a second adversarial pass insisted on — promising that nothing was written. That promise is what the refusal group means, and a middlebox is free to forward the request and only then answer 405 on its own policy, so a caller who skipped reconciliation on that word would leave behind a standing refusal the user believes never took effect; the guard pins exactly that shape, with a double that writes the rule and *then* answers 405, and with the engine's own bare code still landing in the refusal group beside it. (The same passes caught the naive status-only reading and the asymmetry where an empty-string code fell through to a different bucket.) The documented recovery for an unreadable outcome — retry, then reconcile — is pinned to be **ledger-safe** rather than merely asserted: against a double that models the engine's own "is this identity already standing?" question, re-sending the same identity comes back as a no-op with the approval ledger and the bucket revision both unmoved, however many times it is repeated, while two concurrent writers each landing a causal point are both honestly reported as having written. Finally the four local gates are pinned to be free: a missing write verb on the injected port, an unwritable direction, an unreadable identity pair, and a principal key that is present but cannot name anyone all refuse **without sending anything** — the last of those because silently degrading a blank target into absence would land a tightening aimed at someone else in the caller's own bucket and return a 200 |
 | `scripts/run-registrar-tables-test.mjs` | The four registrar table bodies — this Guards table and the three census tables in the repository's negative-control record — are **generated** from `scripts/gates-manifest.json`, the one file that describes a suite. Every row's text must equal what the generator emits, the manifest's suite set must equal the suites on disk, each entry must declare how it is negative-controlled (rehearsed, blind, or behavioural, with the census taker itself declared as such since it does not appear in its own tables), and each row's outward prose is scanned against the published-surface word list — the same list the packaging-hygiene guard uses, shared rather than copied — before the generator may write it into this file. Adding a guard is therefore one manifest entry plus one generator run instead of six hand edits across three files, and a description that drifts in one place and not the others stops being expressible. The row count is no longer what is compared: the earlier arrangement checked the census tables by **length**, so rows naming the wrong suites reconciled green. Positive controls run entirely on in-memory copies — a changed description, a dropped entry, an added entry, a changed class and a hand-edited row on disk each have to make the same judgement speak — and the quieter halves are pinned too: nothing outside a table body may move, a line inside one that is not a recognisable row makes the generator refuse rather than drop it, byte equality is backed by a column-count check (a cell holding a bare pipe splits a row into extra columns, and a code span does not protect it), and the malformed rows kept byte-for-byte as they are found are registered individually, so the registration turns red the day it stops being needed rather than outliving its reason — audited in both directions, since a registration pointing at a row that is no longer malformed and one pointing at a guard that was reclassified or deleted are both exemptions nobody reads |
 | `scripts/run-mcp-probe-face-test.mjs` | The engine's **run-free MCP status face** (server ≥7.93.0, sdk 11.2.0): the `capabilities.mcpProbe` bit read the same four-state way as its eleven sibling capability readers, and two call ports on top of it — read the deployment's own declared servers, or probe a caller-supplied list. Presence of the bit is carried by the engine version, so an **absent key means an older engine** (that route answers a coded 404) and is read as *not reported*, never as *no*; an explicit `false` is the engine's own no and is reported without inventing a reason (whether caller-supplied declarations are accepted is a different bit's question); a non-boolean is malformed and is dropped rather than folded into a no. The availability verdict answers only *should this call go on the wire*: an explicit no means zero requests, and both kinds of *cannot tell* are sent anyway, so an older engine answers with its own coded refusal instead of being silently skipped. One failure judge serves both ports and asks **provenance before status**: a 4xx, 501 or 503 that carries no machine code proves nothing about who answered and is reported as *no verdict*; twelve coded refusals each get their own arm (three identity codes, one of which sits outside the `auth.` family so a prefix fallback would miss it; three different treatments ride the same 400), and a coded answer this version does not recognise lands in *cannot tell*, never in *this engine has no such face*. Retry-after seconds ride 429 and 503 and are absent rather than 0 when the engine gave none. The 200 body is narrowed through the **same per-row narrower** as the streaming `wiring_manifest.mcp[]` leg and the session panel's replay leg, liveness cell included; on this face rows are paired with the submitted declarations **by index**, so a dropped row makes the whole answer unreadable rather than a half table, a row count that differs from the submitted count is reported as misaligned, and an honest empty `servers: []` is kept apart from *non-empty but nothing readable*. `probedAt` and `ttlSec` pass through untouched — this package mints no freshness verdict — the declaration list is handed over as-is with its length snapshotted once, a list that serialises itself differently from what was counted is refused locally with zero requests, and no port ever retries a dial. |
@@ -425,10 +425,10 @@ guard still cross-checks the table by name).
 | `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`). |
-| `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 a retire-by version no more than three minor lines ahead (it fails once the package reaches 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 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-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 four 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; and names the engine refuses at start-up, which it sent as ordinary names), 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 previous release, checked against that release's published file over tens of thousands of inputs; 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). |
 | `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 only from a request built from a plan, whose not-running notice the host shows; a request built without a plan still carries them on engine-fired events, so the engine rejects the whole request loudly instead of a guard hook silently not running — the gate checks both request bodies against the engine's own hooks schema. Without a plan, every removed hook of that kind on an engine-fired event produces one warning per settings port, event and reason. Malformed entries still pass through for the engine to reject loudly, and passing null where the options object goes behaves like passing nothing; a `plan` option that is not a plan is ignored rather than turning the whole page into nothing, and a plan passed directly in place of the options object is recognised and used. 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) now run on the same engine and are held byte-identical to their previous output over every BMP code unit plus random strings. 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. |
+| `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) now run on the same engine and are held byte-identical to their previous output over every BMP code unit plus random strings. 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. |
 | `scripts/run-ask-survives-posture-test.mjs` | The single posture predicate `askSurvivesPosture(card, facts)` for sessions whose standing mode would otherwise answer approval cards on the user's behalf (bypass-style modes). It reads two facts and returns one of three verdicts. The first is the ask origin stamped on the card: the question tool (`content_question`), an organization rule (`org_rule`), a hook (`hook`), an explicit ask rule (`ask_rule`), an organization policy or rule store that could not be read (`org_unavailable`, `rule_store_unavailable`) and the classifier's hand-off after its denial limit (`denial_limit_fallback`) must still be asked (the engine requires a real person to answer all three) and every other origin this build knows is left to the posture only once the host has also reported that its own ask rules did not match. The second is the host's own reading of its settings ask rules for this call: a positive match must be asked, and a command the host could not fully parse counts as no match. When the host reported no reading, every card outside those seven origins gets `unknown`, because an origin says who asked and not that the user's own ask rules did not match; an origin this build does not know gets `unknown` even after a reported non-match. `unknown` is never an approval: the host falls back to its own settings rules. The guard checks the verdict for every origin word, both with no host reading and with a reported non-match, against an independent table whose word set must equal the package's origin list, so a new upstream word fails the guard until it is classified; it covers the combinations of both facts, malformed inputs (non-boolean readings, empty or non-string origins, prototype keys, a different letter case), the fact that the predicate does not read the stronger bits on the card (those stay with the host's earlier checks), real card requests produced by the live-frame, parked-row and suspended-ask paths, and a closed, frozen verdict shape. Since 0.84.0 the product table itself is also checked against the SDK's runtime list of known origin words, so a word the upstream does not know cannot sit in the table. |
 | `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. |
 | `scripts/run-plan-review-dismissal-test.mjs` | An automatic reopen does not put back a plan-review card the user closed (the first-presentation path neither checks nor clears that record, so a replayed park frame still presents its card). A plan-review card the user dismissed (Esc, abort, or any answer that is not approve or reject) is recorded per session and run at the moment of dismissal, synchronously, before anything queued behind the card can be released; a reopen marked `trigger: 'automatic'` then refuses with `{ reopened: false, dismissedByUser: true }` instead of minting a new card the user's next keystroke would land on, while the user's own next action (`trigger: 'user'`) reopens it and clears the record. The record is keyed by gate instance when the host supplies an instance reader: a new plan gate on the same run is still surfaced, and anything that cannot prove the gate is new (no reader, a failed, empty, thrown or timed-out read) refuses on the conservative side. The asynchronous form re-checks after its reads and before minting — a decision handed over meanwhile (seen by the package, or reported by the host's optional hand-over predicate) answers as "your answer is on its way"; a record that changed meanwhile makes the stale evaluation mint nothing and answer from the current record: another close refuses as the user's close and keeps the newer record, a card already back on screen (the user's own action or a concurrent automatic reopen, waited for within the receipt window and re-read once the wait is over) answers `reopened: true`, and a record that is gone (session change, ledger overflow) answers a plain refusal without `dismissedByUser`; a hand-over predicate that throws refuses with a plain `{ reopened: false }`. Each run has at most one reopen on its way: a reopen that arrives while an earlier one's card is published but not yet settled joins it instead of minting a second card and retiring the first card's answer path. Instance readers are snapshotted when they resolve, so a host that hands over its own live set still gets a new gate recognised; and a decisive-looking host answer note does not clear the record for the very card the package's own responder already judged non-decisive (the label was not on that card). A successful reopen replaces only the record taken before the card was minted (with an on-screen marker, not a deletion), a decisive answer clears it, the per-session ledger is bounded, and a session change clears its bucket. Without `trigger` the reopen answers exactly as before, apart from joining a reopen already on its way. A reopen for a session with no view subscribed to question frames (`onQuestionFrameFor(sessionKey, handler)`; `onQuestionFrame` for the default session) refuses with `{ reopened: false, _sema_noPresentationSurface: true }` — in the synchronous and the receipt form alike, per session key, and also when the view goes away while the asynchronous check is reading — and that answer comes before the user-closed check, at the entry and after the asynchronous read alike, whether or not the read proved a newer gate (after the read, a decision handed over while the read was in progress, and a dismissal record that was replaced or cleared meanwhile, still take precedence over the missing view); a host log or probe port that throws changes none of the reopen answers (the suite runs the same arrangement with a quiet and with a throwing log port and requires the same answer and the same card frames); every other refusal (receipt window lapsed, decision in flight, user-closed card, non-default slot without a delivery path, record gone) keeps its own keys without the flag. The first-presentation call does not check for a view: without one it still answers `true` and the card frame goes nowhere, which the suite records as today's behaviour, not as a guarantee. |
@@ -440,7 +440,7 @@ guard still cross-checks the table by name).
 | `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 retired tool, or a name containing "__" without a protocol prefix, 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: it is compared name by name, in both directions, with an independent reading of the installed engine's own tables (retired names and protocol prefixes) 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 |
+| `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. |
 | `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. |
 | `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. |
@@ -455,17 +455,23 @@ guard still cross-checks the table by name).
 | `scripts/run-submit-refusal-projection-test.mjs` | Four kinds of refusal that mean the request itself needs fixing, read by code with one sentence each. Permission rule lists that are too long or hold a bad entry are named precisely (which list, how many entries against the limit, or which entry by position) for a new request and, with different wording, when parked work cannot continue with the request stored for it; the same sentence is appended to a decision that fails for that reason. A tool declaration the engine refuses at the start of a run is read by its code rather than its HTTP status, so a status that looks like a temporary outage is not presented as one: the sentence says the same request will be refused the same way. A deployment refusing a single-user-only setting is attributed from the request the caller sent (the whole-environment shell setting, bypassPermissions, both, or unknown) and always says nothing ran and the local settings are not at fault. A body-shape refusal exposes its two lists of unrecognized and unsupported keys exactly as sent (only well-formed lists are read; a missing list is not treated as empty), with a three-way check of whether any listed key is a settings path. |
 | `scripts/run-mandated-card-values-test.mjs` | Approval cards that newer servers mark as mandatory for four kinds of question (a deployment command policy, an operator's never-auto list, an exhausted durable budget, and supervisor routing) are built with the server's own code and fed through both card paths: the card carries the mandatory mark and the matching reason word, the shared check reports it as mandatory, no mandate word is invented, and a session-wide allow that the server declines to remember produces the single not-remembered notice. An ordinary approval-list card stays non-mandatory and remembered, and the older card shapes are shown to read the other way. This check needs an installed server package to run and reports itself as skipped otherwise. |
 | `scripts/run-park-failed-reason-projection-test.mjs` | The fail-soft true cause when a parked approval cannot be decided and the engine's own `done{suspended}` terminal is the one handed back: that terminal is still returned as-is (no second terminal), but it carries `_sema_park_failed_reason` — the same sentence, byte for byte, that the durable leg puts on its synthesized terminal — and the result frame carries it through with credentials washed the same way `errors[]` is (value only; same sentence as the durable leg after washing). The key is left off when this host has no decision surface at all (no approval card port, a port installed under another session key, or no question overlay): that sentence's way out is to decide on the card, which such a host never shows. Pinned with a policy-fold refusal on decide, both stall exits, a host-retracted card, credential samples through the full bridge on both legs, the no-decision-surface forms, and the absent forms (user interrupt, malformed values, ordinary terminals). |
-| `scripts/run-tool-history-mismatch-projection-test.mjs` | When a provider rejects every request of a session because a tool call in the conversation history no longer pairs up with its result, the terminal error row carries a machine-readable key naming the kind of mismatch and the first tool call id the provider named; a host that offers the rewind command gets the same recovery sentence CC shows (unless that host names the engine and the engine reports it cannot rewind the conversation), plus a count from the second time in a row, while the print and utility lanes, hosts without that command, and the result frame's error text stay unchanged. Only the run's own failure is read, never assistant text, and only when the provider sentence opens the provider's message: other statuses, other 400s and quoted copies do not match. The count is per session and per run, ignores replays, resets on any other outcome and never overstates when it cannot tell; the row-resolving helper that finds the prompt to go back before is checked for the target, the first-turn case and every undecidable case. The error carrier the package mints for a terminal is the same one the failure line is judged on, so handing it to the edit-and-resend reading makes that reading yield exactly when the line carries the mismatch key, across thirty-one terminal shapes including accessor and inherited fields, governance and busy endings among them; the failure line and the carrier are derived from one reading of the frame, so a field whose answer changes between reads cannot make the line and its key disagree, and an empty error sentence yields neither a line nor a carrier. The check also feeds outcomes built and published by an installed server package when one is available and reports that part as skipped otherwise. |
-| `scripts/run-display-untrusted-baseline-diff-test.mjs` | The credential side of the display outlet (`displayUntrusted` and `washCredentials`), diffed byte for byte against the previous published release: five input pools (seeded random strings, exhaustive shapes around labels, quoted values, userinfo, schemes and invisible units, and URL parameters separated by `;` or `,`) run through fifteen outlets. A secret the previous release masked must stay masked wherever it sits in a credential position; the outlet with only the URL net enabled is judged within that net's own reach. Clean lines from the package's own documents pass through unchanged, and every outlet is idempotent (masking the output again changes nothing). The previous release tarball is supplied through `SEMA_DU_BASELINE_TGZ`; without it the diff sections report as skipped and the run counts as partial, not green. The minimal-shape pool also carries scheme-less userinfo whose password holds a quote or angle bracket after a credential label, and the shapes that once washed differently on a second pass. A full-product grammar pool of 64,512 strings joins the idempotence check, and every pool is also checked for the credential stage reaching its pass cap (it must never do so). The pools now include the bare `token` scheme-word forms and a full `token` grammar product, so the new wash is held to the same idempotence and no-regression bars against the previous release. They also include label and `bearer` / `basic` values that run into a `token` scheme word, a non-ASCII space and an address, with or without a leading character, and quoted values where `token` and an address follow `bearer`. |
+| `scripts/run-tool-history-mismatch-projection-test.mjs` | When a provider rejects every request of a session because a tool call in the conversation history no longer pairs up with its result, the terminal error row carries a machine-readable key naming the kind of mismatch and the first tool call id the provider named; a host that offers the rewind command gets the same recovery sentence CC shows (unless that host names the engine and the engine reports it cannot rewind the conversation), plus a count from the second time in a row, while the print and utility lanes, hosts without that command, and the result frame's error text stay unchanged. Only the run's own failure is read, never assistant text, and only when the provider sentence opens the provider's message: other statuses, other 400s and quoted copies do not match. The count is per session and per run, ignores replays, resets on any other outcome and never overstates when it cannot tell; the row-resolving helper that finds the prompt to go back before is checked for the target, the first-turn case and every undecidable case. The error carrier the package mints for a terminal is the same one the failure line is judged on, so handing it to the edit-and-resend reading makes that reading yield exactly when the line carries the mismatch key, across thirty-one terminal shapes including accessor and inherited fields, governance and busy endings among them; the failure line and the carrier are derived from one reading of the frame, so a field whose answer changes between reads cannot make the line and its key disagree, and an empty error sentence yields neither a line nor a carrier. The key also records whether that row offered the recovery sentence, so a host reads the fact instead of re-deciding it later, and the key reader itself is exported and checked from the package entry. The sentence naming the broken turn, which a host appends once it has resolved that turn, is checked for its two exact forms, cleaning before cutting at sixty code points, linear tag stripping, and byte equality with the shipped terminal wording. On any other provider rejection the same terminal row ends with the edit-and-resend sentence and carries that reading as a key: never together with the mismatch key, never on governance, busy or model-output rows, and every row without one of these changes is byte-identical to the previous release. The check also feeds outcomes built and published by an installed server package when one is available and reports that part as skipped otherwise. The arm of the synthesized terminal row is decided in one function that both the row text and the error carrier read, and its order is pinned: a lock or governance terminal keeps its row even when the error text starts with the model-output sentence. |
+| `scripts/run-display-untrusted-baseline-diff-test.mjs` | The credential side of the display outlet (`displayUntrusted` and `washCredentials`), diffed byte for byte against the previous published release: five input pools (seeded random strings, exhaustive shapes around labels, quoted values, userinfo, schemes and invisible units, and URL parameters separated by `;` or `,`) run through fifteen outlets. A secret the previous release masked must stay masked wherever it sits in a credential position; the outlet with only the URL net enabled is judged within that net's own reach. Clean lines from the package's own documents pass through unchanged, and every outlet is idempotent (masking the output again changes nothing). The previous release tarball is supplied through `SEMA_DU_BASELINE_TGZ`; without it the diff sections report as skipped and the run counts as partial, not green. The minimal-shape pool also carries scheme-less userinfo whose password holds a quote or angle bracket after a credential label, and the shapes that once washed differently on a second pass. A full-product grammar pool of 64,512 strings joins the idempotence check, and every pool is also checked for the credential stage reaching its pass cap (it must never do so). The pools now include the bare `token` scheme-word forms and a full `token` grammar product, so the new wash is held to the same idempotence and no-regression bars against the previous release. They also include label and `bearer` / `basic` values that run into a `token` scheme word, a non-ASCII space and an address, with or without a leading character, and quoted values where `token` and an address follow `bearer`. A YAML block-scalar pool (block headers with every chomping and indentation indicator, comments, raw and JSON-escaped line breaks, non-ASCII and tab indentation, invisible characters at line starts, credentials, control strings and bidi characters inside the block, and seven kinds of line after the block) joins all checks: no secret inside a block may show on any outlet that runs the word net, the previous release must show some on the same pool, and the non-block forms in that pool must stay byte for byte as in the previous release. |
 | `scripts/run-registry-conflict-projection-test.mjs` | How a client tells the two meanings of a registry write conflict apart. The registry answers the same 409 when a write merely lost an optimistic-lock race (re-read, re-base, write once more) and when the space being written was deleted, or deleted and re-created under the same name, while the write was in flight (nothing was written, and a re-read would return the replacement space, so a retry would put an old draft into a space nobody reviewed). The package ships the same three-word classification the newer SDK uses under the same name, plus a predicate for caught errors, because the SDK version this package supports does not have them yet. The guard checks every status and body combination against the newer SDK's own function when that SDK is available, checks the predicate against the newer SDK's own error classes, reads every 409 the registry service mints from its source when that tree is available and replays the resulting bodies, and checks that unreadable bodies never throw and never classify as the retryable kind. The parts that need the newer SDK or the registry source report as skipped when those are absent. |
-| `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). |
+| `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 previous release. |
-| `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, and allow rules with a content pattern are never sent. 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. |
+| `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. |
 | `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, 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 two outcomes produce a sentence (the run carried the opt-out and the engine still captures, or the engine says capture cannot be turned off per session), each with its conclusion first so a truncated status line still reads. The guard pins every one of the 72 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`). The check 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. |
 | `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. |
-| `scripts/run-server-7114-fixture-test.mjs` | Adoption checks against the published server package's real task admission, run on every configured server package (the newest one and each older one in the list), each held to an exact expectation for its generation. From 7.114.0, a tool-event hook matcher that selects a tool's alias but not its current name is refused with a field-invalid error naming both names, in every spelling the server reads (literal, pattern, padded, comma list, pipe list, empty segment), and a matcher that also selects the current name is accepted; 7.113 refuses only the literal and the pipe-list spellings; 7.111 and older accept all nine. A server generation outside the table fails the check instead of being guessed, and with fixtures required each generation band must be covered. This package passes the matcher through byte for byte, the request it assembles gets the same answer from the server as the hand-written one, and its local skip check keys on the matcher as written. |
+| `scripts/run-server-7114-fixture-test.mjs` | Adoption checks against the published server package's real task admission, run on every configured server package (the newest one and each older one in the list), each held to an exact expectation for its generation. From 7.114.0, a tool-event hook matcher that selects a tool's alias but not its current name is refused with a field-invalid error naming both names, in every spelling the server reads (literal, pattern, padded, comma list, pipe list, empty segment), and a matcher that also selects the current name is accepted; 7.113 refuses only the literal and the pipe-list spellings; 7.111 and older accept all nine. A server generation outside the table fails the check instead of being guessed, and with fixtures required each generation band must be covered. This package passes the matcher through byte for byte, the request it assembles gets the same answer from the server as the hand-written one, and its local skip check keys on the matcher as written. From 7.115.0 the refusal additionally carries a machine-readable `field` and a `detail` naming the event, the alias and the current name, with the sentence and the error code unchanged; older generations carry neither. |
+| `scripts/run-steer-refusal-advice-test.mjs` | Refusals of a mid-run steer have one speaker. The root entry's `steerRefusalAdviceOf(error)` returns the steering refusal's safety code and disposition sentence, and `ControlRouter.steer` builds its `ControlSafetyError` from that very call, so a client that steers without the router (and therefore holds the raw SDK error) shows the same words the router would instead of keeping its own sentence table. The guard runs a pool of 180 thrown values (every steering code under ten status shapes, every SDK class name, non-steering errors, throwing getters, proxies, inherited fields and non-objects) through both paths and requires the same code, sentence, message, status and cause on every row; it checks every message against the bytes the previous release produced on the same pool, recorded from that build rather than hand-written, with the four rows whose field reads throw registered as the one intended change: the original error now propagates instead of the getter's. It also pins the error's two new read-only fields (`advice`, `status`), absent on the old two- and three-argument constructor calls and on the cancel 404 and local refusals that carry no disposition; the call-site roster of the three steering legs; that each sentence exists exactly once in the source; and that the open-set sentence splits on the same 4xx line the delivery verdict uses. A last section compares each sentence with the table a client kept, so a change on either side forces a fresh reading |
+| `scripts/run-read-deny-tiers-test.mjs` | The check for the built-in read-deny tier lists and for the validator of the deployment key's comma-separated value. The nine sample inputs agreed with the terminal are pinned (an empty input means removing the key and never a stored empty string; `none` is the explicit empty posture and must stand alone; whitespace around entries is trimmed with the same rule the server uses; an empty entry and an unknown or mis-cased tier are refused, while a tier listed twice is accepted and kept as written, exactly as the server reads it), the remaining shapes are covered (Unicode whitespace, separators other than commas, zero-width characters, every tier alone and all of them in either order), a non-string input never throws and a result never carries an empty string as its value, and the explanation sentence is one line, free of internal wording, and shows a user-supplied value only after credential masking, control-character escaping and a length cap (a two-million-character entry is neither echoed nor slow). The validator is also run against a fake tier table: adding a tier makes it accepted, removing one makes it refused, and the source holds no tier name, so the generated table is the only source of tiers. There is no comparison against the server because the server package offers no callable parser for this value. |
+| `scripts/run-engine-caps-settled-within-test.mjs` | The bounded wait for the engine's capability report. `engineCapsSettled` takes an optional `withinMs` and answers whether the wait ended because the probe settled or because the time ran out; without `withinMs` it behaves exactly as before. The bounded form creates a timer only when a probe is actually in flight, clears it on every path, never rejects, treats an unusable budget (zero, negative, not a finite number, a malformed options value) as no waiting rather than waiting forever, clamps an oversized budget to the platform maximum, and counts the budget once per call even if the engine is replaced during the wait. `disableSkillsDeclarationSettled` waits within a bound for the report and then answers exactly as the synchronous `disableSkillsDeclaration` does, so when the probe was started by the host, lands in time and reports the capability as available, the very first request can turn skills off, and in every other case (probe too slow, failed, never started, capability not reported or reported as unavailable) the caller gets the same answer the synchronous mouth gives. It never starts a probe itself. Timing cells run on an injected clock, and a real-timer pair checks that no timer handle is left behind. |
+| `scripts/run-hook-matcher-verdict-test.mjs` | The hook-matcher admission verdict `hookMatcherVerdict(event, matcher)` against the published server package's real task admission: for 35 matcher shapes on each tool event, a `refused` verdict (the matcher selects a tool alias but not that alias's canonical name) matches a 400 from the server, and `accepted` / `matchesAll` match an admitted request; from server 7.115.0 the refusal's machine-readable `detail` (event, alias, canonical) must equal the verdict's pair. Shapes the verdict does not judge are listed with their reason and are never treated as accepted. The guard also pins that the verdict reads the generated alias table (adding a pair to a copy of the table changes the copy's verdict), that only the three tool events are judged, that the verdict is case-sensitive and never throws on non-string input, and that pathological user regexes return within a bounded time (each call runs in a worker with a deadline; the same regex run directly against the longest tool name does not finish). After the fact, `hookMatcherRefusalFromError` reads the server's own refusal: from 7.115.0 the refusal's `field` and `detail` give the event, the alias and the current name, which must equal this function's verdict on the same matcher; older refusals carry no such fields and the reader answers nothing rather than parsing the sentence. |
+| `scripts/run-human-turn-encoding-test.mjs` | The check for the data envelope that carries a user's @-attached file, notebook, image, PDF or directory to an engine that has no attachment seat, and for the single function that turns one human submission into what the request carries. The envelope bytes are pinned against the 45 reference samples agreed with the terminal (the sample file is a byte-for-byte copy), and four extra samples with an instruction inside the file contents are checked for shape: the instruction stays between the two fence lines with a line-number prefix, the fence is longer than any tilde run in the envelope, and the header carries no reminder tag, no claim that a tool was called and no imperative. The three boundary rules (fence length over the whole envelope including look-alike tildes, visible escapes for in-line break characters, quoting and escaping of names) each have their own cells; a malformed descriptor throws and accessor properties are never run. For the assembly: the projection rule (non-empty items joined by a newline, attachments before the item text), the image seat plan (pasted images first, at most 20 images, 6 MiB each and in total, only when the model can see images), the 8 MiB request-body accounting (exactly at the limit is accepted and one byte more degrades or refuses, compared with the real serialized body on six body shapes), the order in which attachments are reduced to a reference sentence, the three-step handling on the follow-up carrier (contents, then reference sentences, then keep queued) and its agreement with the engine's own follow-up text validator on a seeded pool, the delivery shares and the reduced list as two views of one record, the refusal reasons, the image-vision verdict and the attachment-seat capability reader. The engine limits the package mirrors are checked against a server package fixture's real admission path when one is configured. Pasted images take seats first but are held to the same count, per-image and total limits; an image over a limit is left out and recorded as not sent. Very long rider lists are handed to the ledger without overflowing the call stack, and an encoded submission handed to the ledger as-is records partly sent entries as partly delivered. |
+| `scripts/run-server-7115-fixture-test.mjs` | Checks against the published 7.115.0 server package and every older package in the list, each held to an exact per-generation table. A server tool call nobody has spoken for is classified as the new `unstated` ask class by the server's own engine and frame reader, and that pair reaches this package's card and seat unchanged, with the standing-answer verdict matching the engine's own arbitration row. The capabilities route really runs: from 7.115.0 it reports the allow-rule capability and this package reads it as present, older servers do not report it and the allow bucket is not sent; requests this package assembles from a mixed rule pool are accepted by the same generation's admission and settings compiler, while content-form allow rules sent raw are refused there. The server's refusal for tasks parked before an upgrade is produced by its own function and, through the SDK error mapping, is read by this package's dedicated reader and turned into a three-part sentence that says the work is still waiting, with the way out chosen by the parked kind the caller supplies; refusals without the flag are not claimed. The two continue-only-an-existing-conversation refusals are read the same way on every generation, and the prose probe reason minted for live frames equals the durable row's copy and lands on the card from both legs. |
 
 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**