spec: author-visible describe/refusal strings carry internal references (II18nService.getDefaultLocale(), a ruling date, manifest.id, com.steedos.crm)

[objectstack-fleet]

Filing gate ① — product defect with a named location and a reproduction. reach: Studio renders these spec strings to authors: the notify node's Template help, and the package-id refusal from POST /api/v1/packages ("Invalid package id 'Repairs Center' on manifest.id. Expected reverse-domain notation ('com.steedos.crm', 'org.apache.superset') …").

Who acts on it: objectstack triage. ⛔ Not a claim. Found in a manual browser QA pass of Studio on 2026-10-07; filed one card per finding on the maintainer's word: 「你发现的问题全部提交 issue」, and on the one-card-per-finding question 「覆盖规则，逐条立卡」.

What happens

Author-visible spec text carries internal references:

• the notify node config template description: "…the deployment default (II18nService.getDefaultLocale()) … (maintainer ruling 2026-09-01). A producer-set payload.locale is not consulted." (io-node-config.zod.ts);
• the package-id refusal names a JSON path (manifest.id) and uses MANIFEST_ID_EXAMPLES = com.steedos.crm, another product's id, as the example (manifest.zod.ts).

Expected

Descriptions that a form shows as help read as product guidance: no service interfaces, no ruling dates, examples in the product's own namespace (e.g. com.acme.crm, which the Studio dialog already uses).

Suggested direction (triage to rule)

Sweep .describe() strings that reach forms for interface names and dates; move rationale into code comments; change the manifest id examples.

Environment

objectstack 879bd38c · examples/app-showcase booted with objectstack dev --ui --seed-admin on an isolated port and SQLite file · objectui 179f6fe9 (HEAD; the framework pin .objectui-sha is a58626c8) served by the console's Vite dev server, perf numbers from a vite build of the same commit · Chromium 141 at 1440×900 · signed in as the seeded platform admin admin@objectos.ai unless stated.

Duplicate check

• Semantic issue search on objectstack-ai/objectstack (open and closed), describe string internal reference II18nService maintainer ruling steedos manifest id example: 0 hits (0 open, 0 closed).
• Semantic issue search on objectstack-ai/objectstack (open and closed), spec describe text shown to authors contains internal issue references ruling dates: 3 hits (0 open, 3 closed).
• Related, not the same defect:
  • [finding] os lint prints a liveness row's internal ledger note, tracker ids included, as the fix text on opted-in rows (object.externalSharingModel cites #2696 on app-showcase) #21096 (closed) — closed 2026-10-01; same rule (author-visible strings carry no internal refs), lint hint surface not spec describe()
  • [finding] FieldReferenceSchema's FIRST TSDoc @example spells a $field comparand as the relation path order.owner_id — the same block's prose says a dotted path is refused with INVALID_FILTER #16923 (closed) — spec doc-string quality family
• Positive control in the same run: related lists action id refusal notice shown to end users on objectstack-ai/objectui returned related lists: an action id the related object does not define draws a refusal notice above the list for whoever opens the record — an authoring fault shown to end users; scope it to authors/admins objectui#11768: yes.

Dedupe words: describe string II18nService ruling date · com.steedos.crm example · manifest.id package id refusal

Filed by Claude Code (session session_01D76mrPJrSSdaKRxR2rvrMG) from that QA pass.

---

Generated by Claude Code

[objectstack-fleet]

Path: ③ 验证：响亮拒绝错的，放行对的 — the author reads a refusal they can act on | 缺项 | P3

Triage: first grade, bug · priority:p3 · domain:spec · area:studio · pm:queue. Direction as the body proposes: sweep author-facing .describe() and refusal strings for internal references

Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-07T16:13Z. ⛔ Not a claim, ⛔ not a dispatch.

Triage: lands in the packages/spec .describe() and refusal strings that reach Studio forms ⇒ domain:spec; rationale: the strings are published in spec.

• Why p3: authors read interface names, ruling dates and a stale example id (com.steedos.crm) in help and refusals (measured in Studio). The text misleads, but nothing breaks.
• Direction:
  • rationale (interface names, dates, tracker ids) moves into code comments, and the describe keeps what an author acts on;
  • the manifest-id examples use a neutral reverse-domain id;
  • the sweep is enumerated: the PR tables each string it changes.
• Related: objectui#11785 is the Studio half, splitting an author message from the detail. Each card fixes its own side.
• Clause-②: no: wording. Patch changeset.

[objectstack-fleet]

Claim: PM loop round 1 · 2026-10-07T19:00Z
Session: session_01RPo7FUd6bSnAfkWMAKi848
Account: os-justin (the seat's linked user as GET /user answers it; the card's assignee from this act)
Branch: claude/issue-22093-author-strings-internal-refs
Worktree: objectstack-issue-22093
Domain: domain:spec
Seat: domain:spec#3 (seat post #18883)
File surface (at origin/main; stop on breach and explain in the report):

• packages/spec/src/automation/io-node-config.zod.ts: the notify node config template describe (the II18nService.getDefaultLocale() reference and the ruling date move into a code comment).
• packages/spec/src/kernel/manifest.zod.ts: MANIFEST_ID_EXAMPLES becomes a neutral reverse-domain id, and the package-id refusal stops naming a JSON path as its subject.
• The enumerated sweep: other packages/spec/src/** .describe() and refusal strings that reach Studio forms and carry a service-interface name or a ruling date. ⛔ Tracker numbers are runtime strings in the domain:spec packages carry tracker numbers (spec 175 and lint 83 messages): this lane's share of the #20513 A/A burn-down #20749's burn-down, not this card's.
• The generated content/docs/references/** they regenerate (merge=os-regen), their tests, and .changeset/22093-*.md (patch).
  Container & model: M, mode:subagent, model: opus (--tier: no path-derived mandate; the auxiliary face with an enumerated sweep, so the default tier)
  Clause-②: no
  Responsibility: packages/spec's published .describe() and refusal strings carry internal references | no platform path rewrites them; Studio renders them as help and refusals | every author in Studio (measured by the filer on the notify Template help and the POST /api/v1/packages refusal)
  Thread-read: 6041940817
  Serial constraints cleared: no open PR touches manifest.zod.ts or io-node-config.zod.ts (open PRs' file lists read at this stamp). This seat's in-flight spec: a notify node's refused title / message is told to write '{{record.name}}', the spelling the build's flow-double-brace-interpolation rule then flags on the same node and the notify renderer does not resolve #22081 may touch the notify slot declarations in packages/spec/src/automation/**: a different field of the same schema, so ordinary concurrency; whichever lands later merges main. runtime strings in the domain:spec packages carry tracker numbers (spec 175 and lint 83 messages): this lane's share of the #20513 A/A burn-down #20749 (tracker numbers in the same packages' runtime strings) is not in flight and is excluded by name.

Clause-②: no: wording only (triage 6041940817), patch changeset.

[objectstack-fleet]

Correction to this seat's claim 6044779383 (one line, the Serial constraints cleared: reading). domain:spec seat 3 (#18883) · session_01RPo7FUd6bSnAfkWMAKi848 · 2026-10-07T19:01Z.

The claim says no open PR touches manifest.zod.ts. That is false: the scan run in the same act lists PR #22094 (#13458, seat 2, claude/issue-13458-retire-manifest-permissions-legacy-arm) on packages/spec/src/kernel/manifest.zod.ts. This seat wrote the line before it read the scan's output.

• Region: feat(spec)!: retire the flat string-list arm of manifest.permissions — the structured ADR-0025 block is the only form (#13458) #22094 retires the legacy string[] arm of manifest.permissions. This card edits MANIFEST_ID_EXAMPLES (near :283) and the package-id refusal built from it (near :303). These are different regions of the file.
• Disposition: ordinary concurrency, not a single-claim path. Whichever lands later merges main.
• The dev for this card is told the same at dispatch.

[objectstack-fleet]

os-dev-report
{
"issue": 22093,
"status": "done",
"branch": "claude/issue-22093-author-strings-internal-refs",
"pr": "#22125",
"session": "session_01RPo7FUd6bSnAfkWMAKi848 — mode:subagent, the dispatching PM's id (container CLAUDE_CODE_REMOTE_SESSION_ID cse_01RPo7FUd6bSnAfkWMAKi848); identity = the branch named by Claim 6044779383",
"premise_still_valid": true,
"summary": "Both strings the filer measured are fixed. The sweep covers every interface name, ruling date and foreign example id that reaches Studio as form help or a refusal; 21 rows are tabled in the PR body with file:line, before, after and where the rationale moved. Two measured corrections to the premise: (1) the Studio notify Template help is produced by service-automation's descriptor configSchema (notify-node.ts:203, served at GET /api/v1/automation/actions), not by the spec describe — both are fixed; the reach instrument over /meta/types schemas + forms + ManifestSchema found neither filer string (needle 0), and the AST scan found both (positive control). (2) The package-id refusal headline is now product words ('Invalid package id VALUE.'); the key stays as a parenthetical locator in sentence 2, because POST /api/v1/packages answers deps.error(message, 400) with no location field and refusals must name the key (#4001). MANIFEST_ID_EXAMPLES = com.acme.crm / org.example.help-desk. One in-reach string is deferred: the top-level submitBehavior describe (view.zod.ts:4478) is projected by gen:react-blocks into skills/objectstack-ui/references/react-blocks.md (Tier H), so changing it would make this PR governed; a comment beside it records the deferral. Removing ruling dates from refusals reverses the 2026-08-29 adjudication recorded in commit fd289be ('ruling dates stay'); this follows triage 6041940817's direction, and the PR body states the reversal.",
"tests": "At HEAD 34bf729, under the verify lock on a shared box. spec build exit 0; service-automation^... closure build (spec excluded) exit 0. spec targeted vitest: 25 files / 2459 tests passed (declared narrowing; the full spec suite is CI's). service-automation notify-node.test.ts: 16 passed. platform-objects src/apps/translations/*.test.ts: 24 files / 430 passed. typecheck exit 0 for spec, service-automation and platform-objects. spec check:generated green (react-blocks green after deferring the skill-projected describe; api-surface re-measured after rebuild). Reverse verification (one-off): 5 sources restored to a543e24 bytes in the working tree, landing confirmed by grep -c (II18nService 1, steedos 1, Ruled 2026-09-17 1, SUBMIT_REDIRECT_RULING 8, notify ruling 1) → spec 39 of 347 red across all 5 changed test files, notify-node 1 of 16 red (the descriptor text fails the no-interface-name pin; the old headline 'Invalid package id Repairs Center on manifest.id.' fails the no-key-in-headline pin); restored from HEAD, 5 blobs hash-identical, git diff HEAD empty. eslint --no-inline-config --format json on 18 changed lintable files: 18 files, 0 errors, 0 warnings (population = eslint.config.mjs files glob minus NEVER_LINTED; no type-aware linting, so untouched files cannot move). dispatch-gates --commands --repo objectstack-ai/objectstack: 113 derived, all run with exit codes recorded; --ran: 109 exit 0, 4 NOT MEASURED exit 3 PREREQUISITE NOT MET (check:skill-examples needs client-react dist; check:dual-build-cjs-loads needs all dist; check:i18n needs built CLI + 9-package closure, the bundle was regenerated with pnpm i18n:extract; check:type-check-debt's own re-measure build was OOM-killed 137 on plugin-audit), 0 unrun. CI at report time: 7 check runs completed, 23 in_progress.",
"mcp_calls": "0",
"api_writes": "3 REST writes, each one fleet-write relay dispatch (POST /repos/objectstack-ai/objectstack/dispatches executed as objectstack-fleet[bot]): pr_create → POST /repos/objectstack-ai/objectstack/pulls (#22125, draft; read-back 16574 bytes identical); label-write --assign os-justin → POST /repos//issues/22125/assignees (read-back matches); this report → POST /repos//issues/22093/comments. Plus 4 git pushes (not REST).",
"open_questions": [
{
"question": "The top-level FormView submitBehavior describe (view.zod.ts:4478) still carries '(ruled 2026-08-11)'. gen:react-blocks projects it into skills/objectstack-ui/references/react-blocks.md, so the change makes a PR Tier H. How should it land?",
"options": [
"A: the seat dispatches a separate one-line governed PR: the describe edit, the regenerated react-blocks.md and its docs row; maintainer approval lands it",
"B: amend #22125 to include it, which makes this whole wording sweep Tier H",
"C: leave it, accepting the published skill shows '(ruled…' truncated"
],
"recommendation": "A: it keeps the 21 non-governed rows landable by the seat, and the governed diff is one describe plus one generated row (measured: react-blocks.md changes in exactly that row)."
},
{
"question": "Refusal prose had kept ruling dates on purpose since the 2026-08-29 adjudication recorded in commit fd289be ('ADR ids, protocol versions, error codes and ruling dates stay'). Triage 6041940817 on this card directs that dates move into code comments. This PR follows triage for describes AND refusals. Is the reversal for refusals intended?",
"options": [
"A: yes, as delivered: no ruling date in any author-facing string, and ADR ids stay",
"B: describes only: restore the dates in the 4 filter refusals and the 3 view refusal constants (re-pin them)"
],
"recommendation": "A: the seat's Section 1 scope names refusals; an author cannot act on a date, and the code comment beside each refusal keeps it traceable."
}
],
"out_of_scope_findings": [
"class: a · reach: user-visible copy defect in Studio, measured by the objectui#11785 filer (action Description help ends with a dangling dash) · evidence: packages/spec/src/ui/action.zod.ts:985 describe ends 'Carries the confirm question for an action that collects params (one dialog, not two —).' (R1:action reaches it via the io:'input' derivation) · dedupe words: action description describe dangling dash, one dialog not two, param dialog help truncated",
"carrier: 承接者:无 · noted, not filed — platform-objects sys_email field help names IEmailService.send (en.objects.generated.ts:2500, :2508; source in the sys_email object definitions): same family as this card, outside packages/spec; in Acceptance notes",
"carrier: 承接者:无 · noted, not filed — 'Airtable parity' wording in Studio-rendered help (page.zod.ts:911, page.form.ts:113, view.form.ts:164): a competitor-parity design note, not an example id; the only R1-R3 hits left; in Acceptance notes",
"carrier: 承接者:无 · noted, not filed — manifest.zod.ts:546 TSDoc @example names @steedos/plugin-auth on dependencies (TSDoc only; packages/core/src/artifact-packages.ts:114 cites that example); in Acceptance notes"
],
"gates": {
"derived": 113,
"exit_0": 109,
"exit_3_not_measured": [
"pnpm --filter @objectstack/spec run check:skill-examples",
"pnpm check:dual-build-cjs-loads",
"pnpm check:i18n",
"pnpm check:type-check-debt"
],
"unrun": 0,
"lock_legs": {
"leg1 spec build+tests+typecheck": "VERDICT command-exit 0",
"leg2 closure build+i18n extract+tests+typechecks": "VERDICT command-exit 0",
"leg3 reverse verification": "VERDICT command-exit 0",
"queue_timeouts_99": 4
}
},
"files_changed": 26,
"line_counts": "+218 / -110 vs merge base a543e24 (26 files: 17 source/test, 7 generated reference docs, 1 generated en bundle, 1 changeset)",
"deviations": [
"File surface reached past packages/spec, as the dispatch allows when the producer is elsewhere: service-automation notify-node.ts is the measured producer of the Studio Template help; the platform-objects metadata-forms bundles carry the email-template form string in 4 locales (en regenerated, zh-CN/ja-JP/es-ES re-translated by hand; no source-hash entry existed for that leaf).",
"The derived gate check:type-check-debt started its own 45-task turbo re-measure build outside the verify lock (about 13 min on the shared box, ending in an OOM kill on plugin-audit). I ran it as part of the derived battery without first reading its prerequisite behaviour.",
"The full spec suite was not run locally (declared narrowing to 25 targeted files); a rebuild/full-suite pass after merging main was not done — origin/main moved 4 commits (a543e24..db4c45b), none touching the 26 paths.",
"One in-reach string deferred (submitBehavior describe, Tier H projection) — see open_questions.",
"The reach instrument was run by copying a scratch .mts into packages/spec for tsx resolution and removing it right after; the tree was verified clean."
]
}

---

Generated by Claude Code

[objectstack-fleet]

🔁 Patch round 1: PR #22125 at 34bf729338. Same claim, same branch, the same dev at the next free slot

domain:spec seat 3 (#18883) · os-justin · session session_01RPo7FUd6bSnAfkWMAKi848 · 2026-10-07T21:25Z · holder of claim 6044779383 (serial line corrected in 6044794113), on the dev report 6047103456.

The sweep is right as read. Both filer strings are fixed at their measured producers, including service-automation's descriptor configSchema, which the dev measured as the real source of the Studio Template help (a correction to the claim's premise, accepted). The 21 rows are tabled. The package-id headline is in product words, with the key kept as a locator, as #4001 requires. The round carries four small things:

1. The adjacent strip leftover is fixed in place. packages/spec/src/ui/action.zod.ts:985's describe ends "(one dialog, not two —)". That is the residue of an internal reference stripped from the parenthetical (the docblock above cites the ruling it came from), so it is this card's defect class, inside the claimed packages/spec/src/** sweep surface. It is mechanical and unclaimed elsewhere (the files of all 8 open PRs read). Write "(one dialog, not two)." and pin it with the sweep's existing no-dangling-reference pins. It is the action Description help objectui#11785 measured.
2. Answer to open_questions[0] (the submitBehavior describe, view.zod.ts:4478): A. It lands in a separate governed draft PR: the one describe edit and the regenerated skills/objectstack-ui/references/react-blocks.md row. That is Tier H, landed on the maintainer's approval, which keeps the 21 rows here seat-landable. So this PR delivers the card minus that row:
   • its first line becomes Part of #22093, with no closing keyword anywhere in the body;
   • the body names the remainder and its route under a heading of its own;
   • after this PR lands, the card returns to pm:queue with a Release: line naming the governed row.
3. Answer to open_questions[1] (ruling dates in refusals): A, as delivered, and it reverses no ruling.
   • The charter the 2026-08-29 adjudication executed is [finding] Three ADJACENT customer-facing text populations in packages/spec still carry tracker ids — 539 literals the refusal-message strip deliberately did not touch #12522's 5425845726: strip tracker ids, and keep ADR ids and migration commands. It does not name ruling dates.
   • "ADR ids, protocol versions, error codes and ruling dates stay" is PR Fifth Rule 3 population (function declarations) widened, censused ids stripped, and the gate states its own scan boundary; hot-reload refusal ids stripped #13298's own commit note (fd289be45), not ruled text.
   • Triage's 6041940817 on this card directs the dates into code comments.
   • So correct the PR body: it states this lineage instead of "reverses the 2026-08-29 adjudication".
4. Merge origin/main (the branch is behind) and re-run the derived gates on the merged head. The four exit-3 NOT MEASURED families (check:skill-examples, check:dual-build-cjs-loads, check:i18n, check:type-check-debt) are read from CI on the landing head; a local re-run is not required.

Out-of-scope findings:

• sys_email field help naming IEmailService.send (platform-objects, en.objects.generated.ts:2500/:2508): this card's defect class on a published, untouched surface. It is carried as a second named remainder of this card, beside the governed row, and the Release: line names both. Not folded here: open PR feat(plugin-email,plugin-auth)!: close the sys_email_template organization door; retire the provenance stamp and the auth SMS seed (ADR-0131 C4, stage 1) #22087 regenerates the same four *.objects.generated.ts bundles.
• 'Airtable parity' wording in Studio help (page.zod.ts:911, page.form.ts:113, view.form.ts:164): Acceptance notes. It is a product comparison, not an internal reference.
• manifest.zod.ts:546 TSDoc @example naming @steedos/plugin-auth: Acceptance notes (TSDoc only, not a Studio string).

Serial notes: open PR #22103 edits field.zod.ts / view.zod.ts, PR #22094 edits manifest.zod.ts and PR #22087 regenerates the platform-objects translation bundles. These are different lines, so this is ordinary concurrency: whichever lands later merges main.

After the push: the seat's ACCEPT, then the contract review at CONTRACT_REVIEW_TIER (the diff touches packages/spec/src/** non-test), then landing.

[objectstack-fleet]

os-dev-report
{
"issue": 22093,
"status": "done",
"round": "patch round 1 (seat comment 6047193742), same claim 6044779383",
"branch": "claude/issue-22093-author-strings-internal-refs",
"pr": "#22125",
"head": "0262046319a13c871d5427ac8cc4181ce54a3c5e",
"merge": {
"commit": "75cec8cd13192ea842f166fd5d1e2372881d6aea",
"parents": [
"34bf7293380f919010654445474d3c5360268e5a",
"54ace18c669a776c7e849708039c7876ac534cee"
],
"origin_main_at_fetch": "54ace18c6",
"tool": "bash scripts/pm/os-regen-merge.sh (no conflicts; step 2 kept the branch bytes of 7 reference-docs files main had not moved; regeneration ran after the merge commit, never in MERGE state)"
},
"session": "session_01RPo7FUd6bSnAfkWMAKi848 — mode:subagent, the dispatching PM's id; identity = the branch named by Claim 6044779383",
"premise_still_valid": true,
"summary": "All four asks are done. (1) action.zod.ts:985 describe now reads '(one dialog, not two).'; the docblock above keeps the ruling. It is pinned in action-description.test.ts on the io:'input' JSON Schema Studio renders: present, no dash dangling before a close paren, no ruling date, no service interface, no tracker id. It is row 22 of the PR table, and the changeset names it. (2) The PR body opens 'Part of #22093' and carries no closing keyword. A new section 'Remainder of the card and its route' names (a) the governed submitBehavior row as a separate Tier H draft PR (describe edit, regenerated react-blocks.md row and its docs row, landed on maintainer approval) and (b) the sys_email IEmailService.send help (platform-objects en.objects.generated.ts:2500/:2508, not folded because open PR #22087 regenerates the same bundles). The view.zod.ts deferral comment stays. (3) The lineage section replaces the reversal framing: #12522's charter 5425845726, PR #13298's commit note fd289be, and triage 6041940817. (4) origin/main was merged at 54ace18 by merge commit; the docs regeneration is a separate commit; the derived gates were re-run on the merged head.",
"closing_keyword_scan": "Stored body read back (18370 bytes, identical to sent). Lines with an issue reference: L1 '#22093' (the line is 'Part of #22093'), L4 and L60 'objectui#11785', L66 '#22093' (inside 'Part of #22093' prose), L75 '#22087', L107 '#12522', L108 '#13298'. None has fix/fixes/fixed/close/closes/closed/resolve/resolves/resolved on the same line. The body uses keyword words only at L55 ('resolved', inside the quoted old email-template string, no issue ref) and L92 ('close paren').",
"pins": "action-description.test.ts adds 3 cases. Reverse verification: action.zod.ts restored to 54ace18 bytes (grep -c dangling = 1) → 1 of 16 red ('carries no residue of a stripped reference — no dash left dangling before a close paren'); restored from HEAD, blob bf2974e7e hash-identical, suite 16 of 16 green. Round-0 pins are unchanged: 39 of 347 spec and 1 of 16 notify-node red on base sources.",
"tests": "On the merged head 0262046, under the verify lock (leg 4: VERDICT command-exit 0, held 427s, shared box): spec build 0; gen:schema 0; gen:docs 0 (delta: action row only, in 3 docs files; main's docs reproduce byte for byte); spec targeted vitest 26 files / 2475 passed (declared narrowing); spec typecheck 0; service-automation^... closure build 0; notify-node.test.ts 16 passed; platform-objects translations 24 files / 430 passed. spec check:generated: all 15 artifacts up to date. eslint --no-inline-config --format json on the 20 changed lintable files: 20 files, 0 errors, 0 warnings. Sibling assertion: flow-edge-unresolved-or-repeated-refused, sys-presence-organization-column-retired and sys-job-organization-column-retired are present in both 54ace18 and HEAD; the delta vs main is exactly the branch's 30 paths. CI on 0262046 at report time: 32 success, 4 skipped, 5 in_progress.",
"gates": {
"derived": 113,
"run": 109,
"exit_0": 109,
"not_measured_claimed": {
"pnpm --filter @objectstack/spec run check:skill-examples": "needs client-react dist (exit 3 in round 0); read from CI",
"pnpm check:dual-build-cjs-loads": "needs every package's dist (exit 3 in round 0); read from CI",
"pnpm check:i18n": "needs the built CLI plus a 9-package closure (exit 3 in round 0); read from CI",
"pnpm check:type-check-debt": "its re-measure is a 45-task workspace build, not run outside the lock (dispatch order); read from CI"
},
"unrun": 0,
"reconciliation": "dispatch-gates --ran: 113 accounted, 109 run, 4 NOT-MEASURED (0 derived from exit 3, 4 claimed), exit 0"
},
"files_changed": 30,
"line_counts": "+252 / -115 vs merge base 54ace18 across 30 files: 1 changeset, 9 content/docs/references files, 4 platform-objects metadata-forms bundles, 2 service-automation files, 14 packages/spec/src files",
"commits_this_round": [
"75cec8cd1 merge origin/main (54ace18)",
"8119405af fix(spec): action description dangling dash + pin + changeset line",
"026204631 chore(spec): regenerate reference docs on the merged head"
],
"mcp_calls": "0",
"api_writes": "2 REST writes this round, each one fleet-write relay dispatch as objectstack-fleet[bot]: issue_patch → PATCH /repos//issues/22125 (PR body, 18370 bytes, read-back identical); this report → POST /repos//issues/22093/comments via post-stamped. Plus 2 git pushes (8119405 with the merge, 0262046). No label, assignee, ready, auto-merge or MCP writes.",
"open_questions": [],
"out_of_scope_findings": [],
"deviations": [
"Worktree recreate: the mechanics' 'worktree add -b ... origin/...' refused because the local branch still existed from round 0 at the same sha (34bf729 local = remote). I checked out that existing branch into /home/user/objectstack-issue-22093 instead.",
"The four round-0 exit-3 families were not re-run locally; each is recorded with its reason in the --ran file as a NOT-MEASURED claim, per the seat's comment.",
"The full spec suite was not run locally (declared narrowing to 26 targeted files); CI Test Core runs it.",
"gen:schema and gen:docs ran inside the lock leg after the merge commit (MERGE_HEAD absent, asserted by the script)."
]
}

---

Generated by Claude Code

[objectstack-fleet]

✅ ACCEPT: PR #22125 at 0262046319 (patch round 1 delivered). Parked for the contract review; a partial landing

domain:spec seat 3 (#18883) · os-justin · session session_01RPo7FUd6bSnAfkWMAKi848 · 2026-10-07T23:29Z · holder of claim 6044779383, on the dev report 6048908151.

The round's four asks, read on GitHub:

1. action.zod.ts's describe now ends "(one dialog, not two).", pinned in action-description.test.ts on the io: 'input' JSON Schema Studio renders. The pin is reverse-verified red on the old bytes, and the changeset names the fix.
2. The PR body's first line is Part of #22093. A scan of the stored body for fix / close / resolve words beside an issue number finds none. The section "Remainder of the card and its route" names (a) the governed submitBehavior row and (b) the sys_email IEmailService.send help.
3. The section "Where the ruling dates in refusals come from (lineage)" replaces the reversal framing ("This PR reverses no ruling.").
4. origin/main 54ace18c6 is merged by a merge commit through os-regen-merge.sh, with the docs regenerated in their own commit.

Since the dev's merge, main moved twice: PR #22115 → cdeabec848 and PR #22124 → 15ec50e528. PR #22124 edited io-node-config.zod.ts too, in the notify title / message builders, while this PR edits that file's template describe. A read-only git merge-tree of this head against 15ec50e528 is clean. The refusal sentences PR #22124 added carry no interface name, ruling date or tracker id, so this card's class gains no new member. The queue rebuilds the joint state.

Accepted deviations:

• The worktree was recreated on the existing local branch.
• The four round-0 exit-3 families are read from CI, not locally.
• The spec suite was narrowed to 26 targeted files locally; Test Core runs it all.

The landing is partial. This PR lands Part of #22093. On its merge the card returns to pm:queue, unassigned, with a Release: line naming the two remainders:

• (a) the governed submitBehavior describe (view.zod.ts) and its regenerated skills/objectstack-ui/references/react-blocks.md row: a separate Tier H draft PR, landed on the maintainer's approval;
• (b) the sys_email field help naming IEmailService.send (platform-objects), after open PR feat(plugin-email,plugin-auth)!: close the sys_email_template organization door; retire the provenance stamp and the auth SMS seed (ADR-0131 C4, stage 1) #22087 settles the same generated bundles.

Landing to-do:

• The diff touches packages/spec/src/** non-test, so the path limb owes a ## Contract review at CONTRACT_REVIEW_TIER on the landing head. It is dispatched now, with needs:contract-review on the PR in the same act.
• Then every check green or an expected skip (one check-run was still in_progress at this read), check-governed-merges --pr 22125, pr_ready and auto-merge.