mandrel 2.54.0 → 2.56.0

package/.agents/agents/story-worker.md

@@ -48,20 +48,19 @@ the step-by-step. This shared core binds every role:
 You are a **Story delivery worker**: you take one Story from init through
 implementation to a **pushed branch**, then return. You do **not** close it —
 your caller owns the close-and-land tail. Follow the `helpers/deliver-story`
-prose your caller hands you; this delta states the non-negotiable
-MUSTs. Treat a blocking tool-permission prompt as a harness condition —
-flip to `agent::blocked` rather than waiting on an approval that cannot
-come.
+prose your caller hands you; this delta states the MUSTs. Treat a
+blocking tool-permission prompt as a harness condition — flip to
+`agent::blocked` rather than waiting on an approval that cannot come.
 
 ## Worktree discipline (MUST)
 
 1. Initialize with
    `node .agents/scripts/single-story-init.js --story <storyId>` from the
-   **main checkout**, synchronously at max Bash timeout — a per-worktree
-   install can take minutes; do not background it.
+   **main checkout**, synchronously at max Bash timeout — the install
+   can take minutes; never background it.
 2. Capture `workCwd` and `dependenciesInstalled` from the envelope.
-   Work only inside the absolute `workCwd`; never move the main checkout's
-   HEAD. cwd may reset between calls, so anchor every path at `workCwd`.
+   Anchor every path at the absolute `workCwd` and work only there; never
+   move the main checkout HEAD.
 
 ## Verify branch before every commit (MUST)
 
@@ -81,8 +80,8 @@ follow-up commit, never amend.
 ## Docs context — digest first
 
 Do **not** re-read every file in `project.docsContextFiles`. Read the
-`docsDigestPath` digest your caller passes, then pull files on demand at
-the lines it names. A null `docsDigestPath` means no mandate.
+digest your caller passes, then pull files on demand at the lines it
+names. A null digest path means no docs mandate.
 
 ## Close gates — one credited run
 
@@ -90,7 +89,7 @@ the lines it names. A null `docsDigestPath` means no mandate.
 (**typecheck, lint, test, format, maintainability, coverage, crap**) and is
 the authoritative gate — do not pre-run it. The **one** exception is
 the full suite: run it exactly once, after the self-eval loop's last fix
-commit and immediately before the push, in the shape close credits. A bare
+commit and immediately after the push, in the shape close credits. A bare
 `npm test` / `pnpm run test` deposits **no** credit:
 
 ```bash
@@ -102,8 +101,9 @@ node <main-repo>/.agents/scripts/evidence-gate.js --standalone \
 ```
 
 Dispatch it in the **background**: it routinely outruns the host's sync
-Bash ceiling, and its completion re-invokes you — that is the signal. Never spawn a task to poll or `sleep`-loop
-against it; a waiter whose condition is wrong outlives the agent. Share
+Bash ceiling, and its completion re-invokes you — which is why the push
+comes first. Never spawn a task to poll or `sleep`-loop
+against it; a waiter with a wrong condition outlives the agent. Share
 `lint` / `typecheck` evidence with close via `evidence-gate.js`; never
 stamp coverage / CRAP fresh any other way.
 
@@ -117,13 +117,12 @@ Waiter traps: [`parallel-tooling.md`](../workflows/helpers/parallel-tooling.md)
 
 ## Acceptance self-eval before close (MUST)
 
-After the implementation commits land and **before** flipping to `closing`,
-run the bounded acceptance self-eval loop
+**Before** flipping to `closing`, run the bounded self-eval loop
 ([`acceptance-self-eval.md`](../workflows/helpers/acceptance-self-eval.md)).
 It scores the change set you computed **once** and injected into the critic
 — never one it re-derives — against each `acceptance[]` item,
 consuming `verify[]` output as evidence. **proceed** → flip to `closing`,
-push, hand off; **redraft** → fix the flagged criteria, commit, re-eval;
+push, capture, hand off; **redraft** → fix the criteria, commit, re-eval;
 **block** → take the blocked path below. Never hand off an unscored
 branch.
 
@@ -141,20 +140,22 @@ branch.
 
 The init envelope carries `remoteVerified` + `remoteProbe`. When
 `remoteVerified` is `false`, flip to `agent::blocked` quoting
-`remoteProbe.detail` and stop. A PR opened by
-`single-story-close.js` is the only sanctioned landing.
+`remoteProbe.detail` and stop. A PR opened by `single-story-close.js` is
+the only sanctioned landing.
 
 ## Your turn ends at a pushed branch (MUST)
 
 You do **not** run close. Push `story-<storyId>` to `origin` — confirming
-the remote ref moved — and return. The dispatching orchestrator runs
+the remote ref moved — **before** the credited capture, then return: the
+capture is backgrounded, so its completion ends your turn, and a turn that
+ends unpushed reads as unfinished work. The orchestrator runs
 `single-story-close.js` in its own session, serialized against your
-siblings. Do not open the PR, flip `agent::done`, or spawn a child to close
-on your behalf. If the push fails, take the blocked path above.
+siblings. Do not open the PR, flip `agent::done`, or spawn a child to
+close for you. If the push fails, take the blocked path above.
 
 ## Return contract — the hand-off report
 
 A short, literal hand-off your caller can act on: Story id, `workCwd`,
-branch, pushed head SHA, self-eval verdict, `verify[]` evidence. Say plainly
-the branch is pushed and unclosed. Never hand-compose a terminal envelope —
+branch, pushed head SHA, self-eval verdict, `verify[]` evidence. Say the
+branch is pushed and unclosed. Never hand-compose a terminal envelope —
 inventing one makes an unlanded Story look landed.

package/.agents/audit-checklists/accessibility.md

@@ -24,7 +24,4 @@ Self-check your change against this lens's concerns before you ship:
 - [ ] Media alternatives
 - [ ] Contrast where statically derivable
 - [ ] Raw-element census
-- [ ] Resolve the target from config — never a hardcoded URL.
-- [ ] Sample routes from the navigability SSOT.
 - [ ] Run an accessibility engine per sampled route.
-- [ ] Median-of-3 or provisional.

package/.agents/audit-checklists/mobile.md

@@ -28,8 +28,4 @@ Self-check your change against this lens's concerns before you ship:
 - [ ] Typography and spacing
 - [ ] Coverage — is anything exercised at a non-desktop viewport?
 - [ ] Effectiveness — does that exercise assert anything mobile-specific?
-- [ ] Resolve the target from config — never a hardcoded URL.
-- [ ] Sample routes from the navigability SSOT.
 - [ ] Drive two form factors per route.
-- [ ] Median-of-3 or provisional.
-- [ ] Leave the viewport as you found it.

package/.agents/docs/agentrc-reference.json

@@ -28,6 +28,10 @@
     "projectOwner": null,
     "operatorHandle": "@[USERNAME]",
     "defaultTimeoutMs": 60000,
+    "followUpRepos": {
+      "framework": "dsj1984/mandrel",
+      "platform": null
+    },
     "branchProtection": {
       "enforce": true,
       "requiredChecks": [
@@ -75,7 +79,8 @@
     ],
     "memoryPool": {
       "staleAfterDays": 30,
-      "growthDelta": 25
+      "growthDelta": 25,
+      "indexByteCeiling": 24576
     },
     "failOnSharedEditors": false,
     "requireExplicitCrossStoryDeps": false,
@@ -353,7 +358,8 @@
       },
       "autoMerge": "trust-ci",
       "blockOnAdvisoryFailure": true,
-      "advisoryAllowlist": []
+      "advisoryAllowlist": [],
+      "rerunAdvisory": 0
     },
     "routing": {
       "roleScopedAgents": true,

package/.agents/docs/configuration.md

@@ -103,6 +103,9 @@ GitHub provider identity plus the remote stance the bootstrap enforces. `owner`,
 | `projectOwner` | No | `string` \| `null` | `null` | Owner of the Projects V2 board when it lives outside `owner` (an org board fed by a user repo). `null` means the board shares `owner`. |
 | `operatorHandle` | Yes | `string` | `"@[USERNAME]"` | The human the framework escalates to, `@`-prefixed. Used for HITL @-mentions on `agent::blocked`. |
 | `defaultTimeoutMs` | No | `integer` | `60000` | Default `timeoutMs` applied to every `gh` subprocess the provider facade spawns, so a stalled socket or long-poll cannot hang an orchestration indefinitely. A `GhExecTimeoutError` from a hit ceiling is classified `transient` and retried by `withTransientRetry`. Story #2860. |
+| `followUpRepos` | No | `object` | — | Repository slugs for the non-consumer follow-up ownership buckets, used when a CI gap, retro proposal, or audit finding belongs to someone other than the repo that surfaced it. |
+| `followUpRepos.framework` | No | `string` | `"dsj1984/mandrel"` | `<owner>/<repo>` that owns framework-level defects. Defaults to the Mandrel mirror — the one bucket with a knowable default. |
+| `followUpRepos.platform` | No | `string` \| `null` | `null` | `<owner>/<repo>` for a shared platform or infrastructure tracker (a shared base config, a runner fleet, a cross-repo toolchain). No default — nothing can guess a shared repo. Left unset, platform-owned findings file locally and say so. |
 | `branchProtection` | No | `object` | — | Branch-protection stance applied to `project.baseBranch` by the GitHub bootstrap, and reproduced locally before every push. |
 | `branchProtection.enforce` | No | `boolean` | `true` | When true, the GitHub bootstrap writes the required-check ruleset. False leaves the remote stance alone. |
 | `branchProtection.requiredChecks[]` | No | `array<object>` | `[{"name":"lint","cmd":["npm","run","lint"]},{"name":"test","cmd":["npm","test"]},{"name":"baselines","cmd":["node",".agents/scripts/check-baselines.js"]}]` | Checks that must pass before a Story PR merges. Each entry carries both the remote context name and the local argv. Each item has: name, cmd. |
@@ -130,6 +133,7 @@ Inputs to `/mandrel-plan`: risk escalation heuristics, ceremony-lite routing, th
 | `memoryPool` | No | `object` | — | Thresholds for the memory-hygiene advisory `/mandrel-plan` surfaces at Gate #1. Advisory only: it recommends `/memory-consolidate` and never gates, reroutes, or mutates the memory pool. |
 | `memoryPool.staleAfterDays` | No | `integer` | `30` | Recommend a consolidation pass once the pool's stamp is older than this many days. Default 30. |
 | `memoryPool.growthDelta` | No | `integer` | `25` | Recommend a consolidation pass once this many entries have been written since the last one. Measured against the entry count the last pass stamped, so a stamp predating that field leaves growth unmeasured and only the age threshold applies. Default 25. |
+| `memoryPool.indexByteCeiling` | No | `integer` | `24576` | Recommend a consolidation pass once the pool's `MEMORY.md` index exceeds this many bytes. Independent of the age and growth thresholds: the harness truncates the index it loads into each session at its own byte cap, so an oversized index is a loss already happening — every entry listed after the cut is invisible — rather than a hygiene forecast. Default 24576, the harness cap itself. |
 | `failOnSharedEditors` | No | `boolean` | `false` | When true, upgrade shared-editor conflict findings to hard errors (default false — advisory soft findings only). |
 | `requireExplicitCrossStoryDeps` | No | `boolean` | `false` | When true, upgrade implicit cross-Story dependency findings to hard errors (default false — advisory soft findings only). |
 | `crossCuttingRegistries` | No | `string[]` or `{ append?, prepend? }` | `["lib/orchestration/lifecycle/listeners/index.js","**/listeners/index.js","**/handlers/index.js"]` | Registry path patterns whose concurrent edits across Stories are flagged as conflicts. Defaults to the framework listener/handler index patterns when omitted. |
@@ -330,6 +334,7 @@ Everything `/mandrel-deliver` and `single-story-close` consume: execution timeou
 | `ci.autoMerge` | No | `"trust-ci"` \| `"strict"` | `"trust-ci"` | Story #4356 (Epic #4355). Merge posture. 'trust-ci' (default) merges once required checks pass; 'strict' additionally requires a clean review gate. |
 | `ci.blockOnAdvisoryFailure` | No | `boolean` | `true` | Story #5096. When true (default), delivery refuses to arm — and disarms — GitHub native auto-merge while a non-required (advisory) check is genuinely red on the PR head and GitHub reports the PR mergeable anyway (mergeStateStatus=UNSTABLE). `--auto` waits on REQUIRED contexts only, so without this a red advisory quality gate merges unattended. Set false to restore the pre-#5096 behaviour verbatim. |
 | `ci.advisoryAllowlist` | No | `array<string>` | `[]` | Story #5096. Check-run names exempt from blockOnAdvisoryFailure — a red run whose name matches exactly never blocks arming. Matching is exact; an unnamed run can never match and always blocks. |
+| `ci.rerunAdvisory` | No | `integer` | `0` | Story #5266. How many times close may re-run a failed advisory workflow run before blocking on it, per close invocation. Default 0: close spends no CI minutes and issues no GitHub mutation on an advisory red unless asked. At n > 0 the failed run(s) are re-run within that allowance and the merge wait re-polls inside its existing budget, landing or blocking on the re-run verdict. Overridden per invocation by --rerun-advisory <n>. |
 | `routing` | No | `object` | — | v2 delivery-spawn routing: role-scoped boot contexts and maker-checker sampling. The v1 singleDelivery epic-route kill-switch was removed in Stage 6. |
 | `routing.roleScopedAgents` | No | `boolean` | `true` | Epic #4478 (M7-B). Kill-switch for the role-scoped boot contexts. When true (default), a converted delivery spawn (`story-worker`, `acceptance-critic`) boots on its own `.claude/agents/<role>.md` system prompt instead of re-paying the full CLAUDE.md @-import closure. When false, every converted spawn falls back to `subagent_type: general-purpose` — the instant, code-rollback-free per-consumer revert, and the universal escape for hosts that ignore `.claude/agents/`. The fallback is the full-closure agent that ran before M7-B, so flipping it off never drops a gate. |
 | `routing.freshCriticSampleRate` | No | `number` | `0.2` | Epic #4478 (M7-B, Part 2). Maker-checker sampling floor. Under the standard profile, a change set touching no sensitive path routes its acceptance clusters down the contract-identical inline critic path, but this fraction of them is still forced through a fresh-context critic so a low derived level never means zero independent checking. Clamped to [0, 1]; 0 disables the floor, 1 forces every cluster fresh. Consumed by resolveCeremonyForRisk (lib/orchestration/ceremony-routing.js). |

package/.agents/rules/ci-remediation.md

@@ -27,13 +27,29 @@ exactly one of two ways, and no others:
    through the fix table in
    [`deliver-story-reference.md` § Step 4](../workflows/helpers/deliver-story-reference.md#step-4--ci-watch--fix-recovery);
    refresh a baseline only when the diff demonstrably can't be covered.
-2. **File a `meta::framework-gap` issue** when the root cause is outside this
+2. **File the CI-gap intake issue** when the root cause is outside this
    delivery's scope — a pre-existing flaky test, a runner/infra weakness, a
-   framework-level environment gap. Open the issue with the `meta::framework-gap`
-   label (see [`git-conventions.md`](git-conventions.md)) carrying **the run
-   link and the failure signature** so a later `/mandrel-plan` Phase 0 sweep can act on
-   it. Remediate this delivery only if the pre-existing defect is genuinely
-   blocking it.
+   framework-level environment gap. One command does it, and it is the only
+   sanctioned filing surface:
+
+   ```bash
+   node .agents/scripts/file-ci-gap.js --story <id> --verdict <verdict> \
+     --owner <consumer|framework|platform> --evidence "<proof reading>" [--block]
+   ```
+
+   It reads the digest for the run link and failure signature, routes the
+   filing to the repository that **owns** the fault, dedups by signature so
+   the Nth occurrence updates the existing ticket, posts the `friction`
+   comment, and (with `--block`) flips the Story. Hand-running `gh issue
+   create` is not the fallback: it files an issue no `/mandrel-plan` pass can
+   graduate, in whichever repo you happen to be standing in. Remediate this
+   delivery only if the pre-existing defect is genuinely blocking it.
+
+   **`--owner` is the judgement call**, and it is yours to make from the
+   evidence: `consumer` for this repository's own code, `framework` for a
+   Mandrel defect, `platform` for a shared base config, runner fleet, or
+   cross-repo toolchain that neither owns. An unconfigured bucket files
+   locally and says so in the issue body — it never pretends to be routed.
 
 Infra, transient, and flaky failures are root-cause defects too — a flaky test
 that passes on a rerun is still a bug that will fail a future run. They route
@@ -49,9 +65,9 @@ the two options above. Name the verdict you reached in the `friction` comment.
 | Verdict | Evidence | Routes to |
 | --- | --- | --- |
 | **defect-in-diff** | The failure reproduces on the branch and not on an unmodified `main` | Option 1 — fix at source |
-| **pre-existing** | The same check fails on an unmodified `main` too | Option 2 — file `meta::framework-gap`; remediate here only if it blocks this delivery |
-| **capacity** | Proven exhaustion of a runner resource, not a property of the diff (see below) | Option 2 — file `meta::framework-gap` **and** escalate to the operator |
-| **unreproducible-tier** | The tier cannot be exercised in this sandbox at all, proven by an attempted attach (see below) | Option 2 — file `meta::framework-gap` **and** escalate on first encounter |
+| **pre-existing** | The same check fails on an unmodified `main` too | Option 2 — `file-ci-gap.js --verdict pre-existing`; remediate here only if it blocks this delivery |
+| **capacity** | Proven exhaustion of a runner resource, not a property of the diff (see below) | Option 2 — `file-ci-gap.js --verdict capacity` (`meta::framework-gap` unless `--owner` routes it elsewhere) **and** escalate to the operator |
+| **unreproducible-tier** | The tier cannot be exercised in this sandbox at all, proven by an attempted attach (see below) | Option 2 — `file-ci-gap.js --verdict unreproducible-tier` (`meta::framework-gap` unless `--owner` routes it elsewhere) **and** escalate on first encounter |
 
 Why the verdict set carries these last two is recorded in
 [`docs/decisions.md` ADR 20260906-5160a](../../docs/decisions.md).
@@ -72,10 +88,12 @@ line naming the exhausted limit (an OOM kill, `ENOSPC`, `EMFILE`,
 timeout), plus the fact that the failure is not specific to this diff. Absent
 that reading the verdict is **flaky, not capacity**, and it routes to Option 1.
 
-On a `capacity` verdict: file the `meta::framework-gap` issue with the run link,
-the failure signature, and the resource reading; flip the Story to
-`agent::blocked` with a `friction` comment naming the verdict; and hand back to
-the operator, who owns the runner pool. Do not sit in a retry loop waiting for
+On a `capacity` verdict: run `file-ci-gap.js --verdict capacity --block`, passing
+the resource reading as `--evidence` (the run link and failure signature come
+from the digest). That files the intake issue — `meta::framework-gap`, or
+`meta::platform-gap` when `--owner platform` names a shared runner fleet — posts
+the `friction` comment and flips the Story in one call; then hand back to the
+operator, who owns the runner pool. Do not sit in a retry loop waiting for
 capacity to return.
 
 **Rerunning a failed job to reach green stays forbidden under every verdict,
@@ -106,10 +124,9 @@ both:
   failure in the app under test.
 
 Absent both readings the verdict is unavailable and the failure routes as it did
-before. On the verdict: file the `meta::framework-gap` issue with the run link,
-the failure signature, and the attach attempt; flip the Story to
-`agent::blocked` with a `friction` comment naming the verdict; and hand back to
-the operator, who owns the sandbox. Do not author a fix for a tier you could not
+before. On the verdict: run
+`file-ci-gap.js --verdict unreproducible-tier --block`, passing the failed attach
+as `--evidence`; then hand back to the operator, who owns the sandbox. Do not author a fix for a tier you could not
 run — a blind fix to a suite nobody exercised is how the gap compounds.
 
 ## Verifier
@@ -131,8 +148,9 @@ alongside the failing check-run identity. On green it adjudicates:
 
 - **Same head SHA** → the green came from re-running the failed job. The
   watcher exits non-zero, flips the Story to `agent::blocked` with a
-  `friction` comment, and requires the `meta::framework-gap` issue (run link +
-  failure signature, both already in the digest) before the delivery proceeds.
+  `friction` comment, and requires the CI-gap intake issue
+  (`file-ci-gap.js` — run link and failure signature are already in the
+  digest) before the delivery proceeds.
 - **New head SHA** → fix at source. The digest is retired, auto-merge is
   re-armed, and the delivery continues unobstructed.
 
@@ -153,8 +171,8 @@ operator under **any** of:
 - **Clearly-environmental → escalate immediately.** An unambiguously
   environmental failure outside your control (runner provisioning, a persistent
   registry/network outage, a branch-protection or CI misconfiguration, an
-  expired credential) — file the `meta::framework-gap` issue (with run link +
-  signature) and escalate on the first encounter rather than burning iterations
+  expired credential) — run `file-ci-gap.js --block` and escalate on the
+  first encounter rather than burning iterations
   trying to code around it. A proven-capacity failure is this case: reach the
   `capacity` verdict above and escalate on the first encounter.
 - **Unrunnable tier → escalate immediately.** A tier the sandbox cannot host at

package/.agents/schemas/agentrc.schema.json

@@ -169,6 +169,25 @@
           "description": "Default `timeoutMs` applied to every `gh` subprocess the provider facade spawns, so a stalled socket or long-poll cannot hang an orchestration indefinitely. A `GhExecTimeoutError` from a hit ceiling is classified `transient` and retried by `withTransientRetry`. Story #2860.",
           "default": 60000
         },
+        "followUpRepos": {
+          "type": "object",
+          "description": "Repository slugs for the non-consumer follow-up ownership buckets, used when a CI gap, retro proposal, or audit finding belongs to someone other than the repo that surfaced it.",
+          "properties": {
+            "framework": {
+              "type": "string",
+              "pattern": "^[^/\\s]+/[^/\\s]+$",
+              "description": "`<owner>/<repo>` that owns framework-level defects. Defaults to the Mandrel mirror — the one bucket with a knowable default.",
+              "default": "dsj1984/mandrel"
+            },
+            "platform": {
+              "type": ["string", "null"],
+              "pattern": "^[^/\\s]+/[^/\\s]+$",
+              "description": "`<owner>/<repo>` for a shared platform or infrastructure tracker (a shared base config, a runner fleet, a cross-repo toolchain). No default — nothing can guess a shared repo. Left unset, platform-owned findings file locally and say so.",
+              "default": null
+            }
+          },
+          "additionalProperties": false
+        },
         "branchProtection": {
           "type": "object",
           "description": "Branch-protection stance applied to `project.baseBranch` by the GitHub bootstrap, and reproduced locally before every push.",
@@ -377,6 +396,12 @@
               "minimum": 1,
               "description": "Recommend a consolidation pass once this many entries have been written since the last one. Measured against the entry count the last pass stamped, so a stamp predating that field leaves growth unmeasured and only the age threshold applies. Default 25.",
               "default": 25
+            },
+            "indexByteCeiling": {
+              "type": "integer",
+              "minimum": 1,
+              "description": "Recommend a consolidation pass once the pool's `MEMORY.md` index exceeds this many bytes. Independent of the age and growth thresholds: the harness truncates the index it loads into each session at its own byte cap, so an oversized index is a loss already happening — every entry listed after the cut is invisible — rather than a hygiene forecast. Default 24576, the harness cap itself.",
+              "default": 24576
             }
           },
           "additionalProperties": false
@@ -1975,6 +2000,12 @@
               },
               "description": "Story #5096. Check-run names exempt from blockOnAdvisoryFailure — a red run whose name matches exactly never blocks arming. Matching is exact; an unnamed run can never match and always blocks.",
               "default": []
+            },
+            "rerunAdvisory": {
+              "type": "integer",
+              "minimum": 0,
+              "description": "Story #5266. How many times close may re-run a failed advisory workflow run before blocking on it, per close invocation. Default 0: close spends no CI minutes and issues no GitHub mutation on an advisory red unless asked. At n > 0 the failed run(s) are re-run within that allowance and the merge wait re-polls inside its existing budget, landing or blocking on the re-run verdict. Overridden per invocation by --rerun-advisory <n>.",
+              "default": 0
             }
           },
           "additionalProperties": false
@@ -2084,7 +2115,9 @@
                         "not": {
                           "pattern": "([;&|`]|\\$\\()"
                         },
-                        "minLength": 1
+                        "minLength": 1,
+                        "pattern": "^[a-z0-9][a-z0-9._-]*(?:\\/[a-z0-9][a-z0-9._-]*)+$",
+                        "description": "Tier-relative skill id, e.g. `stack/qa/acme-sso`: lowercase segments of letters, digits, `.`, `_` or `-`, at least two of them, separated by `/`. A traversal (`../..`), an absolute path, a backslash or an uppercase segment is rejected here rather than normalized."
                       }
                     },
                     "required": ["skill"],

package/.agents/schemas/lifecycle/merge.unlanded.schema.json

@@ -30,7 +30,8 @@
         "arm-failure",
         "api-race-other",
         "predicate-refused",
-        "advisory-gate-red"
+        "advisory-gate-red",
+        "advisory-gate-inconclusive"
       ]
     },
     "reason": { "type": "string", "minLength": 1 },

package/.agents/schemas/story-deliver-terminal.schema.json

@@ -131,6 +131,7 @@
             "api-race-other",
             "predicate-refused",
             "advisory-gate-red",
+            "advisory-gate-inconclusive",
             "merged-flip-failed"
           ]
         },