`DELIVERY_NOT_ELIGIBLE`'s ledger gloss describes only the `redeliver` refusal, and is now narrower than the code's uses

[os-sam]

Filed by the domain:services PM seat (session session_01APWX2AwT3a4xDcjPCe8bk4) because the contract-review tier ruled it and named this seat as the one to card it. Recording, not claiming — unassigned, and deliberately without pm:queue: routing is not grading.

⚠️ Filed rather than fixed, on purpose. packages/spec has zero ownership in the domain:services lane, so this seat may not take it. It is a one-line docs catch-up for the spec lane.

The gap

packages/spec/src/api/error-code-ledger.zod.ts:511 glosses DELIVERY_NOT_ELIGIBLE in redeliver-flavoured wording — "delivery row is in a non-terminal state". That described the code's only use when it was written.

PR #11858 (card #11453) added a second use on the sibling surface: INotificationOutbox.ack now refuses any row that is not in_flight — which includes pending rows and already-terminal rows. So the code now refuses on both sides of the state space the gloss describes as the refusable one.

⇒ The name is exactly right for both — "this delivery row's state forbids the operation" — and only that one comment lags behind.

Why the fix is the gloss and not the code

The contract-review tier ruled this explicitly (PR #11858's PASS, 2026-08-24 22:39:33Z, tier verified by fuse):

Open question ruled: B — the reuse is right and the ledger's inline gloss for DELIVERY_NOT_ELIGIBLE … is now narrower than the code's uses; that is a one-line spec-lane docs catch-up … not a blocker here, and option C's second near-synonym code is rejected for the report's own vocabulary-sprawl reason.

The rejected alternative was minting a distinct code for the ack refusal. That would need an ERROR_CODE_LEDGER entry plus a row in packages/runtime's dispatcher-error vocabulary, and it would put two near-synonyms in a closed set whose whole purpose is to stay small — the sprawl ADR-0112 exists to prevent. One concept, one spelling, on two delivery surfaces is the ruled shape.

Suggested scope

Widen the gloss so it covers both refusal families rather than naming only the redeliver one. ⚠️ Worth stating in the comment which two surfaces use it, so the next reader does not have to re-derive that the same code serves an HTTP redeliver refusal and a notification ack refusal — that shared vocabulary is real and easy to mistake for a copy-paste.

⛔ Nothing about behaviour changes; nothing in packages/runtime's vocabulary changes; the code name does not change.

Refs

#11453 (the card whose repair half raised it) · PR #11858 (the second use, and the PASS that ruled this) · ADR-0112 (the closed error-code set) · #11009 (the predicate-discard trap the same PR dodged — unrelated to this gloss, listed so the PR's context is findable)

[claude]

Concentrated triage batch: finding → pm:queue, Task, S (domain:spec stands) — the contract-review tier already ruled this needs carding: DELIVERY_NOT_ELIGIBLE's ledger gloss (error-code-ledger.zod.ts:511) still describes only the redeliver refusal while PR #11858 added the ack use (refusing pending and terminal rows). One-line gloss catch-up naming both surfaces. Clause-②: no (gloss prose only) — re-affirm at claim.

---

Generated by Claude Code

[os-litant]

Claim: PM loop round R6
Session: session_01NDGG54XF5gbTLdQzCtnaVV
Branch: claude/issue-11892-delivery-not-eligible-gloss
Worktree: objectstack-issue-11892
Domain: domain:spec
File surface: packages/spec/src/api/error-code-ledger.zod.ts (DELIVERY_NOT_ELIGIBLE gloss lines only) + regenerated ledger docs page + changeset (stop on breach; explain in the report)
Container & model: S text-surface, mode:subagent, model: opus — dispatch-gates 2026-08-25 @0b04839: "no path-derived mandate … The tier stays the PM's per-card judgment call (floor sonnet · default opus · ceiling fable)"; gloss accuracy is editorial judgment no gate checks ⇒ default opus, not sonnet floor
Clause-②: no
Serial constraints cleared: none — sole writer of error-code-ledger.zod.ts in this batch; no registry.ts touch; generated-docs regen disjoint from sibling regens (skills/** belongs to #11930). No pin assertions on sibling behavior.

---

Generated by Claude Code

[os-litant]

{
 "issue": 11892,
 "status": "done",
 "branch": "claude/issue-11892-delivery-not-eligible-gloss",
 "pr": "https://github.com/objectstack-ai/objectstack/pull/12089",
 "premise_still_valid": true,
 "summary": "Widened the ERROR_CODE_LEDGER gloss for DELIVERY_NOT_ELIGIBLE from the single clause 'delivery row is in a non-terminal state' to a per-surface statement covering every refusal a repo-wide grep actually finds. The card's core premise holds and is stronger than filed: the old wording was not merely narrow, it was BACKWARDS for half the code's uses - redeliver refuses NON-terminal rows while INotificationOutbox.ack refuses anything not in_flight, which includes terminal rows, so no single status predicate can gloss both. Comment only: no code registered or removed, no wire value changed, no accept/reject behaviour moved; the legal-metadata set is byte-identical before and after. Patch changeset added because packages/spec publishes src/**/*.zod.ts, so the corrected gloss ships to consumers. No sweep of other ledger entries for the same gloss-drift class was attempted - out of scope for this card and not speculated on.",
 "tests": "All verdicts quoted from each gate's own output line, exit codes captured before any pipe. Union re-run on the FINAL commit 285254d (git rev-parse --short HEAD), after the commit, not before. POST-COMMIT @285254d: pnpm check:error-code-casing -> 'no unlisted lowercase error codes in 4660 scanned file(s) (ADR-0112).' rc=0 | pnpm check:dispatcher-error-vocabulary -> 'check-dispatcher-error-vocabulary: OK - 21 unregistered code-stamping site(s), all classified; 1 awaiting a ledger entry (#8846).' rc=0 | pnpm --filter @objectstack/spec check:docs -> '229 generated files in sync with packages/spec' rc=0 | check:liveness rc=0 ('every governed-type property ... the README state table carries a row for each of the 31 governed type(s)') | check:empty-state rc=0 ('all classified (1 closed, 2 open, 4 output, 9 scope)') | check:strictness-ledger rc=0 ('strictness ledger: 60 file(s) across 5 triaged director(ies)') | pnpm --filter @objectstack/spec typecheck rc=0 ('check:test-typecheck: OK - @objectstack/spec test layer compiles under packages/spec/tsconfig.test.json; 55 file(s) / 263 error(s) held in test-typecheck-debt.json'). SAME TREE, PRE-COMMIT (content byte-identical to 285254d): pnpm --filter @objectstack/spec exec vitest run --maxWorkers=2 src/api/error-code-ledger.test.ts -> 'Test Files 1 passed (1) / Tests 17 passed (17)'; plus 20 more gate families green (list in gates field). NO ablation was performed and none is claimed: this diff removes no guard and adds no gate, so there is nothing whose failure could be demonstrated by mutation - a mutation of prose would only prove that prose changed. The edit was instead confirmed on disk by anchored greps on both the removed and the injected text (old 'delivery row is in a non-terminal state' count 0, new 'does not permit the requested operation' count 1) rather than by any editor exit code, and control-byte scanned (grep -naP over the control range, no match).",
 "gates": {
 "derivation": "node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack (no paths passed; script took its own merge-base change set of 2 files). 26 families matched - BROADER than the 8 in the dispatch prompt, which predated the .changeset file. Header confirmed the answer came from objectstack-ai/objectstack @ f7b25c5.",
 "passed": ["check:error-code-casing", "check:dispatcher-error-vocabulary", "@objectstack/spec check:docs", "@objectstack/spec check:liveness", "@objectstack/spec check:empty-state", "@objectstack/spec check:strictness-ledger", "@objectstack/spec check:variant-docs", "@objectstack/spec typecheck", "@objectstack/spec error-code-ledger.test.ts", "check:changeset-gate-self-tests", "check:cross-package-test-inputs", "check:merge-driver", "check:objectui-changeset", "check:published-files", "check:slot-lookup", "check:spec-parsed-alias", "check:test-source-alias", "check:type-source-resolution", "@objectstack/lint check:doc-formula-expressions", "check-adr-0087-registration", "check-changeset-no-major", "check-ci-filter-parity", "check-cross-package-test-inputs", "check-empty-changeset", "check-plugin-teardown-shape", "check-affected-docs", "check-drift-comment", "release-rehearsal-clone --self-test", "check-nul-bytes"],
 "failed": [],
 "not_run": ["check-dev-prereqs.mjs - PREREQUISITE NOT MET, not a finding: it reported 67 of 67 workspace packages missing their dist/ entry point and exited before measuring anything. Needs a full pnpm build, a repo-scale run CI owns and performs before this gate.", "pnpm lint (repo-wide eslint) - narrowed to the changed files WITH the three-part proof, see deviations."],
 "ci_status": "in_progress - report delivered at draft-PR time per the dispatch contract; CI convergence is the PM's read, not mine. No CI polling was done."
 },
 "line_budget": {
 "applies": false,
 "reason": "No skills/** file in the diff, so the published-skill net-growth budget does not bind. Reported for completeness.",
 "total": "+64 / -1 across 2 files",
 "packages/spec/src/api/error-code-ledger.zod.ts": "+28 / -1 (one 1-line trailing gloss replaced by a 28-line block comment, matching the file's existing style for richly-documented entries such as UPDATE_ID_MISMATCH and ERR_SYSTEM_WRITE_ORGANIZATION_REQUIRED)",
 ".changeset/delivery-not-eligible-gloss-both-surfaces.md": "+36 (new file, patch to @objectstack/spec)"
 },
 "files_changed": [
 "packages/spec/src/api/error-code-ledger.zod.ts",
 ".changeset/delivery-not-eligible-gloss-both-surfaces.md"
 ],
 "deviations": [
 "PM mechanical assumption 1 PARTLY FALSIFIED. The ledger has ONE gloss for this code, not 'at least two occurrences'. git grep over packages/**/*.ts returns exactly two textual hits in error-code-ledger.zod.ts, and the first is the cross-reference INSIDE DELIVERY_NEVER_SENT's gloss on the line above; both sit in the same @objectstack/service-messaging section, so ':513' and 'the @objectstack/service-messaging section' name the same place. Consequence: the PM's suggested per-package-section route had nothing to iterate over - one gloss, one edit. Also line drift: the issue cites :511, the entry is at :513 on f7b25c5.",
 "PM mechanical assumption 2 REFINED UPWARD (code beat the triage reading, as instructed). The triage named two surfaces; grep over the real throw sites found SIX, in two families. redeliver (HttpRedeliverError): (a) row not terminal - assertHttpRedeliverable, http-outbox.ts:350; (b) producer RedeliverGuard returned a refusal - :400; (c) RedeliverGuard itself THREW, fail-closed - :394; (d) terminal re-check at the write missed after a dispatcher tick re-claimed the row - sql-http-outbox.ts:413 and memory-http-outbox.ts:190. ack (NotificationAckError): (e) row not in_flight, covering BOTH unclaimed pending and already-terminal - sql-outbox.ts:227 and memory-outbox.ts:135; (f) compare-and-set read-back shows the claim was lost mid-ack - sql-outbox.ts:298. (b), (c) and the CAS misses are not state predicates at all, which is a SECOND reason no one-clause gloss covers this code - beyond the opposite-halves point the card made. The gloss covers the measured set.",
 "PM mechanical assumption 3: MECHANISM CONFIRMED, CONSEQUENCE FALSIFIED. content/docs/references/api/error-code-ledger.mdx IS an auto-generated artifact (header marker; generator packages/spec/scripts/build-docs.ts, gen:docs / check:docs). But the generator emits only the bulleted CODE-NAME list - it never carries the inline glosses (grep for the old wording in the .mdx returns nothing). So this change yields ZERO docs delta and no generated file belongs in this PR. Proven, not assumed: after gen:schema supplied the gitignored json-schema tree that check:docs requires, check:docs reported '229 generated files in sync with packages/spec' rc=0. gen:schema itself produced no tracked-file change (authorable-surface.base.json untouched; it verified against upstream f7b25c546b64). No MERGE state was present when gen:schema ran (git rev-parse -q --verify MERGE_HEAD -> none).",
 "BASE is f7b25c5, not the PM's 0b04839 - origin/main advanced between dispatch and worktree creation. Branch cut from origin/main at f7b25c5; all readings are against that tree.",
 "Declared narrowing: repo-wide pnpm lint was NOT run; targeted eslint on the changed files was, with all three required pieces of evidence. (1) Population read from eslint's own config, not guessed: eslint itself declines the changeset with 'File ignored because no matching configuration was supplied.', so the linted population of this diff is the single .ts file. (2) Count read from --format json: 2 entries, 0 errors, 0 real warnings (the single 'warning' IS that ignore notice). (3) Invariance for untouched files: this repo runs one eslint.config.mjs that never enables type-aware linting for ANY file - no parserOptions.project, no typed @typescript-eslint rules - stated in the config itself at line ~328 with a measured positive control, so a two-file comment diff cannot move any untouched file's verdict. This is a measurement, not a skip.",
 "Declared narrowing: check-dev-prereqs.mjs was not run to a verdict - its precondition is a fully built workspace (67/67 packages missing dist/ entry points) and it explicitly measured nothing.",
 "@objectstack/lint check:doc-formula-expressions initially exited 1 on BOTH first attempts, each time as PREREQUISITE NOT MET (an unbuilt @objectstack/formula, then an unbuilt @objectstack/lint) - the gate states in its own output that it measured nothing and that this is NOT a finding. Resolved by two targeted single-package builds; the gate then passed with '22 record-scoped formula example(s) across 421 files / 1449 TS blocks judged clean'.",
 "MECHANISM FACT FOR PM, not a blocker here: the additive-label curl recipe in the dispatch contract is NOT reachable from this dev seat. A direct curl POST to api.github.com returned HTTP 403 with 'GitHub access is not enabled for this session. An org admin must connect the Claude GitHub App for this organization.' The gh CLI is also absent ('gh: command not found'). Git push over HTTPS and the GitHub MCP tools both work, and the PR was created through MCP. No skip-changeset label was needed on this PR (it carries a real changeset), so nothing was lost - but a future card that needs that label cannot use the documented curl form from this seat.",
 "The MCP create_pull_request call auto-appends its own session-URL signature footer; my body carried one too, producing a duplicate. Corrected with one body update that omits the manual footer. Noted because it is a channel behaviour, not another actor editing the PR."
 ],
 "open_questions": [],
 "out_of_scope_findings": []
}

---

Generated by Claude Code

[os-litant]

os-dev-report

Supersedes the previous comment: its leading <!-- os-dev-report --> marker was stripped by the body sanitizer on write (verified by read-back — the stored body begins at the fence), which would have made it invisible to a marker scan. Same payload, literal-text marker. No comment-edit endpoint is reachable from this seat, so this is a repost rather than a fix in place.

{
 "issue": 11892,
 "status": "done",
 "branch": "claude/issue-11892-delivery-not-eligible-gloss",
 "pr": "https://github.com/objectstack-ai/objectstack/pull/12089",
 "premise_still_valid": true,
 "summary": "Widened the ERROR_CODE_LEDGER gloss for DELIVERY_NOT_ELIGIBLE from the single clause 'delivery row is in a non-terminal state' to a per-surface statement covering every refusal a repo-wide grep actually finds. The card's core premise holds and is stronger than filed: the old wording was not merely narrow, it was BACKWARDS for half the code's uses - redeliver refuses NON-terminal rows while INotificationOutbox.ack refuses anything not in_flight, which includes terminal rows, so no single status predicate can gloss both. Comment only: no code registered or removed, no wire value changed, no accept/reject behaviour moved; the legal-metadata set is byte-identical before and after. Patch changeset added because packages/spec publishes src/**/*.zod.ts, so the corrected gloss ships to consumers. No sweep of other ledger entries for the same gloss-drift class was attempted - out of scope for this card and not speculated on.",
 "tests": "All verdicts quoted from each gate's own output line, exit codes captured before any pipe. Union re-run on the FINAL commit 285254d (git rev-parse --short HEAD), after the commit, not before. POST-COMMIT @285254d: pnpm check:error-code-casing -> 'no unlisted lowercase error codes in 4660 scanned file(s) (ADR-0112).' rc=0 | pnpm check:dispatcher-error-vocabulary -> 'check-dispatcher-error-vocabulary: OK - 21 unregistered code-stamping site(s), all classified; 1 awaiting a ledger entry (#8846).' rc=0 | pnpm --filter @objectstack/spec check:docs -> '229 generated files in sync with packages/spec' rc=0 | check:liveness rc=0 ('every governed-type property ... the README state table carries a row for each of the 31 governed type(s)') | check:empty-state rc=0 ('all classified (1 closed, 2 open, 4 output, 9 scope)') | check:strictness-ledger rc=0 ('strictness ledger: 60 file(s) across 5 triaged director(ies)') | pnpm --filter @objectstack/spec typecheck rc=0 ('check:test-typecheck: OK - @objectstack/spec test layer compiles under packages/spec/tsconfig.test.json; 55 file(s) / 263 error(s) held in test-typecheck-debt.json'). SAME TREE, PRE-COMMIT (content byte-identical to 285254d): pnpm --filter @objectstack/spec exec vitest run --maxWorkers=2 src/api/error-code-ledger.test.ts -> 'Test Files 1 passed (1) / Tests 17 passed (17)'; plus 20 more gate families green (list in gates field). NO ablation was performed and none is claimed: this diff removes no guard and adds no gate, so there is nothing whose failure could be demonstrated by mutation - a mutation of prose would only prove that prose changed. The edit was instead confirmed on disk by anchored greps on both the removed and the injected text (old 'delivery row is in a non-terminal state' count 0, new 'does not permit the requested operation' count 1) rather than by any editor exit code, and control-byte scanned (grep -naP over the control range, no match).",
 "gates": {
 "derivation": "node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack (no paths passed; script took its own merge-base change set of 2 files). 26 families matched - BROADER than the 8 in the dispatch prompt, which predated the .changeset file. Header confirmed the answer came from objectstack-ai/objectstack @ f7b25c5.",
 "passed": ["check:error-code-casing", "check:dispatcher-error-vocabulary", "@objectstack/spec check:docs", "@objectstack/spec check:liveness", "@objectstack/spec check:empty-state", "@objectstack/spec check:strictness-ledger", "@objectstack/spec check:variant-docs", "@objectstack/spec typecheck", "@objectstack/spec error-code-ledger.test.ts", "check:changeset-gate-self-tests", "check:cross-package-test-inputs", "check:merge-driver", "check:objectui-changeset", "check:published-files", "check:slot-lookup", "check:spec-parsed-alias", "check:test-source-alias", "check:type-source-resolution", "@objectstack/lint check:doc-formula-expressions", "check-adr-0087-registration", "check-changeset-no-major", "check-ci-filter-parity", "check-cross-package-test-inputs", "check-empty-changeset", "check-plugin-teardown-shape", "check-affected-docs", "check-drift-comment", "release-rehearsal-clone --self-test", "check-nul-bytes"],
 "failed": [],
 "not_run": ["check-dev-prereqs.mjs - PREREQUISITE NOT MET, not a finding: it reported 67 of 67 workspace packages missing their dist/ entry point and exited before measuring anything. Needs a full pnpm build, a repo-scale run CI owns and performs before this gate.", "pnpm lint (repo-wide eslint) - narrowed to the changed files WITH the three-part proof, see deviations."],
 "ci_status": "in_progress - report delivered at draft-PR time per the dispatch contract; CI convergence is the PM's read, not mine. No CI polling was done."
 },
 "line_budget": {
 "applies": false,
 "reason": "No skills/** file in the diff, so the published-skill net-growth budget does not bind. Reported for completeness.",
 "total": "+64 / -1 across 2 files",
 "packages/spec/src/api/error-code-ledger.zod.ts": "+28 / -1 (one 1-line trailing gloss replaced by a 28-line block comment, matching the file's existing style for richly-documented entries such as UPDATE_ID_MISMATCH and ERR_SYSTEM_WRITE_ORGANIZATION_REQUIRED)",
 ".changeset/delivery-not-eligible-gloss-both-surfaces.md": "+36 (new file, patch to @objectstack/spec)"
 },
 "files_changed": [
 "packages/spec/src/api/error-code-ledger.zod.ts",
 ".changeset/delivery-not-eligible-gloss-both-surfaces.md"
 ],
 "deviations": [
 "PM mechanical assumption 1 PARTLY FALSIFIED. The ledger has ONE gloss for this code, not 'at least two occurrences'. git grep over packages/**/*.ts returns exactly two textual hits in error-code-ledger.zod.ts, and the first is the cross-reference INSIDE DELIVERY_NEVER_SENT's gloss on the line above; both sit in the same @objectstack/service-messaging section, so ':513' and 'the @objectstack/service-messaging section' name the same place. Consequence: the PM's suggested per-package-section route had nothing to iterate over - one gloss, one edit. Also line drift: the issue cites :511, the entry is at :513 on f7b25c5.",
 "PM mechanical assumption 2 REFINED UPWARD (code beat the triage reading, as instructed). The triage named two surfaces; grep over the real throw sites found SIX, in two families. redeliver (HttpRedeliverError): (a) row not terminal - assertHttpRedeliverable, http-outbox.ts:350; (b) producer RedeliverGuard returned a refusal - :400; (c) RedeliverGuard itself THREW, fail-closed - :394; (d) terminal re-check at the write missed after a dispatcher tick re-claimed the row - sql-http-outbox.ts:413 and memory-http-outbox.ts:190. ack (NotificationAckError): (e) row not in_flight, covering BOTH unclaimed pending and already-terminal - sql-outbox.ts:227 and memory-outbox.ts:135; (f) compare-and-set read-back shows the claim was lost mid-ack - sql-outbox.ts:298. (b), (c) and the CAS misses are not state predicates at all, which is a SECOND reason no one-clause gloss covers this code - beyond the opposite-halves point the card made. The gloss covers the measured set.",
 "PM mechanical assumption 3: MECHANISM CONFIRMED, CONSEQUENCE FALSIFIED. content/docs/references/api/error-code-ledger.mdx IS an auto-generated artifact (header marker; generator packages/spec/scripts/build-docs.ts, gen:docs / check:docs). But the generator emits only the bulleted CODE-NAME list - it never carries the inline glosses (grep for the old wording in the .mdx returns nothing). So this change yields ZERO docs delta and no generated file belongs in this PR. Proven, not assumed: after gen:schema supplied the gitignored json-schema tree that check:docs requires, check:docs reported '229 generated files in sync with packages/spec' rc=0. gen:schema itself produced no tracked-file change (authorable-surface.base.json untouched; it verified against upstream f7b25c546b64). No MERGE state was present when gen:schema ran (git rev-parse -q --verify MERGE_HEAD -> none).",
 "BASE is f7b25c5, not the PM's 0b04839 - origin/main advanced between dispatch and worktree creation. Branch cut from origin/main at f7b25c5; all readings are against that tree.",
 "Declared narrowing: repo-wide pnpm lint was NOT run; targeted eslint on the changed files was, with all three required pieces of evidence. (1) Population read from eslint's own config, not guessed: eslint itself declines the changeset with 'File ignored because no matching configuration was supplied.', so the linted population of this diff is the single .ts file. (2) Count read from --format json: 2 entries, 0 errors, 0 real warnings (the single 'warning' IS that ignore notice). (3) Invariance for untouched files: this repo runs one eslint.config.mjs that never enables type-aware linting for ANY file - no parserOptions.project, no typed @typescript-eslint rules - stated in the config itself at line ~328 with a measured positive control, so a two-file comment diff cannot move any untouched file's verdict. This is a measurement, not a skip.",
 "Declared narrowing: check-dev-prereqs.mjs was not run to a verdict - its precondition is a fully built workspace (67/67 packages missing dist/ entry points) and it explicitly measured nothing.",
 "@objectstack/lint check:doc-formula-expressions initially exited 1 on BOTH first attempts, each time as PREREQUISITE NOT MET (an unbuilt @objectstack/formula, then an unbuilt @objectstack/lint) - the gate states in its own output that it measured nothing and that this is NOT a finding. Resolved by two targeted single-package builds; the gate then passed with '22 record-scoped formula example(s) across 421 files / 1449 TS blocks judged clean'.",
 "MECHANISM FACT FOR PM, not a blocker here: the additive-label curl recipe in the dispatch contract is NOT reachable from this dev seat. A direct curl POST to api.github.com returned HTTP 403 with 'GitHub access is not enabled for this session. An org admin must connect the Claude GitHub App for this organization.' The gh CLI is also absent ('gh: command not found'). Git push over HTTPS and the GitHub MCP tools both work, and the PR was created through MCP. No skip-changeset label was needed on this PR (it carries a real changeset), so nothing was lost - but a future card that needs that label cannot use the documented curl form from this seat.",
 "The MCP create_pull_request call auto-appends its own session-URL signature footer; my body carried one too, producing a duplicate. Corrected with one body update that omits the manual footer. Noted because it is a channel behaviour, not another actor editing the PR.",
 "REPORT-CHANNEL FACT FOR PM: the '<!-- os-dev-report -->' HTML-comment marker did NOT survive the issue-comment sanitizer - confirmed by read-back, the stored body began at the code fence with the marker gone. This report is a repost led by the literal text 'os-dev-report'; the earlier markerless comment (5407595434) carries the identical payload. No comment-edit endpoint is reachable from this seat, so the markerless one cannot be removed or amended."
 ],
 "open_questions": [],
 "out_of_scope_findings": []
}

---

Generated by Claude Code