class (c): a dotted messages id is authorable, schema-valid and resolves to nothing — and TranslationItemSchema's own @example teaches it, at two sites

Surfaced by #17768's dev as a class-(c) out_of_scope_findings entry; the domain:devx seat verified two of its three legs and names the third as unverified. ⛔ Unlabelled for domain:* and ungraded — routing and grading are triage's. ⛔ The filer did not dedupe; dedupe words at the bottom.

Class (c) — authorable metadata that the runtime silently drops

A messages id containing a dot is writable, schema-valid, and resolves to nothing. Nothing refuses it at either door.

Verified by the seat

1 — messages is a FLAT record, so a dotted id is a key that merely contains a dot, ⛔ not a path:

packages/spec/src/system/translation.zod.ts:670
  messages: z.record(z.string(), z.string()).optional()
             .describe('UI message translations keyed by message ID')
control: 163 occurrences of `z.` in that file ⇒ the grep discriminates

2 — the spec's own docblock @example teaches the trap, and it does so TWICE — ⚠️ the dev reported one site; there are two:

packages/spec/src/system/translation.zod.ts:577    "messages": { "common.save": "Save" }
packages/spec/src/system/translation.zod.ts:1430    messages: { 'common.save': '保存' }

⭐ That is the load-bearing part. A docblock @example on the schema is the doorway an authoring agent copies from — so the trap is not merely possible, it is demonstrated to the author by the contract itself. It will keep being minted until this is fixed.

⛔ NOT verified by the seat — establish it before acting

The dev reports that "both t() implementations walk the key segment by segment, so a dotted id is unreachable". The seat did not confirm this. It looked in packages/spec/src/system/i18n-resolver.ts and found the only split('.') there (:3263) is a schema-title helper, unrelated to translation lookup — so the t() implementations are elsewhere and were not located.

⇒ ⛔ Do not treat the unreachability mechanism as measured. Find both t() implementations, show what a dotted id does at each, and say whether the two agree — if one resolves it and the other does not, that is a bigger finding than this card describes.

Corroboration already in the tree

packages/plugins/plugin-audit/src/translations/messages.ts:3-10 is reported to document this rule privately, for its own bundle — i.e. a team already hit it and wrote themselves a local note instead of the rule reaching the contract. Worth reading as evidence of the class, ⛔ not as a second measurement.

Suggested shape (⛔ a proposal, not a prescription)

Two independent halves; ⛔ do not assume one implies the other:

Fix the doorway — the two @example blocks must stop demonstrating an id that cannot resolve. This is cheap and is the half that stops new instances.
Decide whether the door should refuse — a single-segment constraint on messages keys would make the trap unauthorable rather than merely undocumented. ⚠️ That narrows an accept set on a published schema ⇒ it is a Clause-②: yes shape and owes the contract-review path; it is ⛔ not a drive-by addition to the docs fix.

Acceptance

The t() behaviour is measured at both implementations, with a firing control (an id that does resolve).
The @example blocks teach only ids that resolve, verified by running one through the real lookup rather than by eye.
If the schema is narrowed, that rides its own card with a Clause-② declaration — ⛔ not folded into the example fix.

Dedupe words

messages dotted key · translation message id segment · resolveKey dot walk · TranslationItemSchema example · single-segment message id

Generated by Claude Code

Claim: PM loop round 1
Session: session_01JbZnqu8bt6YqfJsr9vaFb3
Branch: claude/issue-18190-messages-example-dotted-id
Worktree: objectstack-issue-18190
Domain: domain:spec
Seat: domain:spec#2 (seat post #18549 — second concurrent seat on this lane; seat 1 is #6017 and this claim does not touch it)
File surface: packages/spec/src/system/translation.zod.ts (the two docblock @example blocks, re-derived this round at :577 and :1430), plus whatever content/docs/references/** page those docblocks regenerate, .changeset/*.md. Read-only: packages/core/src/fallbacks/memory-i18n.ts, packages/spec/src/system/i18n-resolver.ts, packages/plugins/plugin-audit/src/translations/messages.ts, packages/spec/src/type-alias-convention.pin.test.ts (stop on breach; explain in the report)
Container & model: S→M, mode:subagent, model: default judgement tier — quoting this round's own node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --tier packages/spec/src/system/translation.zod.ts: "Model tier — no path-derived mandate: the surface hits none of the 3 declared glob(s)" and "Clause ② SUSPECT surface — a hint, not a verdict"
Clause-②: no
Thread-read: 5706385077
Serial constraints cleared: packages/spec/src/system/translation.zod.ts is held by no in-flight card. ⚠️ The nearest hold is packages/lint/src/validate-translation-references.ts (#18203, PR #18433 enqueued; #18441 and #18442 queued behind it) — same SUBJECT (translation references), different file and different package, and this card's scope is deliberately kept out of the lint rule. If the work turns out to need an edit in packages/lint/**, that is a breach: stop and report, ⛔ do not cross into it. Seat 1's other held files and this seat's own #16894 / #17849 surfaces are disjoint. packages/spec/src/type-alias-convention.pin.test.ts is shared with #16894 and declared READ-ONLY on both.

⚠️ SCOPE — this dispatch takes half 1 ONLY. The card names two independent halves and says in terms 「⛔ do not assume one implies the other」:

✅ IN SCOPE — fix the doorway. The two @example blocks must stop demonstrating an id that cannot resolve.
⛔ OUT OF SCOPE — narrowing the schema. A single-segment constraint on messages keys narrows a published accept set, which the card itself marks as a Clause-②: yes shape owing the contract-review path. It rides its own card. ⛔ Not a drive-by addition here. If the dev believes it must be folded in, that is a fork: stop and report, ⛔ do not widen this PR.

⇒ This claim's Clause-②: no is scoped to half 1 and is only true because half 2 is excluded.

Premise re-derivation, taken 2026-09-17T00:48Z against origin/main — both taught sites are still on the tree:

packages/spec/src/system/translation.zod.ts:577   *   "messages": { "common.save": "Save" }
packages/spec/src/system/translation.zod.ts:1430  *   messages: { 'common.save': '保存' },
control: 163 occurrences of `z.` in that file ⇒ the grep discriminates

⭐ The card's one unverified leg is now measured — but by the TRIAGE seat, not by this seat and not by the dev. Triage comment 5706385077 located resolveKey at packages/core/src/fallbacks/memory-i18n.ts:104-113 and read that it walks key.split('.') segment by segment against a flat z.record(z.string(), z.string()). That is a relayed reading and this seat did not re-take it. The card's own acceptance item 1 demands BOTH implementations be measured with a firing control — triage found one. ⇒ The dev re-takes it first-hand, locates the second implementation, and states whether the two agree; if one resolves a dotted id and the other does not, the card says that is a bigger finding than it describes — report it, ⛔ do not absorb it.

Acceptance item 2 is a measurement, not an edit: the replacement @example ids must be verified by running one through the real lookup, ⛔ 「not by eye」.

Generated by Claude Code

os-dev-report

{
  "issue": 18190,
  "status": "done",
  "branch": "claude/issue-18190-messages-example-dotted-id",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/18553",
  "session": "session_01JbZnqu8bt6YqfJsr9vaFb3",
  "premise_still_valid": true,
  "summary": "Re-took the premise first-hand at origin/main 79a046f8cd: both taught sites are still on the tree (translation.zod.ts:577 JSON @example, :1430 TypeScript @example; control 163 occurrences of 'z.' in that file). Both @example blocks now author the single-segment id commonSave, and both docblocks state the rule, so the contract stops demonstrating an id that cannot resolve. ZONE 1 held: the schema is NOT narrowed - measured leg 0 shows the OLD dotted payload still parses clean at both TranslationItemSchema and TranslationDataSchema, so the published accept set is byte-identical and Clause-2 stays 'no'. A patch changeset is owed and written: the docblock publishes twice over (packages/spec files[] ships both 'dist' and 'src/**/*.zod.ts'; commonSave verified present in dist/translation.zod-BVZ_8ktO.d.ts). No content/docs/references page regenerates from these blocks - those pages carry .describe() property tables only, and check:generated reports all 15 artifacts up to date. The assignee was already set to os-bill by the PM dispatch and was not touched; the newest Claim: comment (5706712291) names this branch.",
  "zone2_mechanism": "MEASURED FIRST-HAND, AND THE SECOND IMPLEMENTATION IS NOW LOCATED. (1) packages/core/src/fallbacks/memory-i18n.ts - resolveKey() at :106 splits on '.' and walks segment by segment; t() at :152 calls it, and again on the declared fallbackLocale leg at :169. (2) packages/services/service-i18n/src/file-i18n-adapter.ts - a SECOND resolveKey() at :27 with the same body; t() at :142 reaches it through resolveFromLocale() at :277, authored layer first then static bundle. THE TWO AGREE: same split, same walk, same undefined at the first missing hop, same 'return the key itself' on a miss. The bigger finding the card warned about - one resolving a dotted id while the other does not - DOES NOT EXIST. Nothing was absorbed into this PR on that account. Corroboration read but not counted as a measurement: plugin-audit/src/translations/messages.ts:3-10 states the rule privately for its own bundle, and content/docs/protocol/kernel/i18n-standard.mdx:493-500 already prescribes exactly the spelling this PR adopts (messages.commonSave) - which is why commonSave was chosen rather than invented.",
  "tests": "All figures at 9620486d10, the branch's final commit; origin/main did not move from the branch point 79a046f8cd, so no merge was owed. ACCEPTANCE ITEM 2 (a measurement, not an edit): a one-time script drove the EXACT authored payload of each repaired @example through BOTH real t() implementations, loaded through their built entry points, via the real authored path (readAuthoredTranslationLayer's sys_metadata row -> envelope strip -> replaceAuthoredTranslations) and via loadTranslations for the JSON example. Exit 0, VERDICT: ALL LEGS AS PREDICTED. BEFORE: t('messages.common.save','zh-CN') returned \"messages.common.save\" at memory-i18n AND at file-i18n-adapter (the key itself = resolves to nothing). AFTER: t('messages.commonSave','zh-CN') returned \"保存\" at both. Firing control t('objects.account.label','zh-CN') returned the translation on every leg, so a silent no-op is excluded. The script lived in packages/runtime (the one package resolving all three), was deleted after the run, and git status is clean - no permanent test file left behind. GATES: dispatch-gates.mjs --repo objectstack-ai/objectstack --commands derived 78 families from the real change set (2 paths, three-dot semantics); all 78 run, exit codes landed to disk BEFORE any pipe, reconciled with --ran carrying 'command :: exit N': '78 derived famil(ies) accounted for - 78 run, 0 NOT-MEASURED (a DERIVED zero - all 78 recorded an exit code and none of them is 3)'. check:i18n-walk-parity (the family PM flagged as load-bearing) exit 0: '11 declared group(s), 8 walked, 3 exempted'. check:generated exit 0: 'All 15 generated artifacts are up to date', check:docs among them. spec typecheck + test exit 0: 483 test files, 13773 tests passed. Repo-wide eslint . --no-inline-config was run IN FULL rather than narrowed: 6803 files, 0 errors, 0 warnings, exit 0. ONE RED, CONTROLLED AND NOT MINE: check:cross-package-test-inputs exit 1, naming packages/cli/test/init-created-files-summary.e2e.test.ts descending packages/spec/dist/. The gate enumerates the real filesystem (readdirSync), so it was ablated: with packages/spec/dist/ moved aside it exits 0 ('OK: 29 package(s) read outside themselves, all declared'); restored, it exits 1 - identical source tree, restore verified by directory presence and no leftover, under a trap. Trigger is that spec was built locally; CI runs this gate on an unbuilt checkout and never sees it. Neither named file is in this diff. Filed below. THREE RESULTS THAT WERE NOT MEASUREMENTS AND WERE RE-RUN: check:doc-formula-expressions, check:dual-build-cjs-loads and check:lean-entry-closure each first answered PREREQUISITE NOT MET (exit 3, 'nothing was measured'), and check-plugin-teardown-shape --self-test refused on a shallow clone (positive control commit unreachable). After building the named packages and git fetch --unshallow, all four exit 0. The reconciliation reflects the re-runs.",
  "gates": "78/78 derived families run and reconciled with exit codes; 0 NOT-MEASURED, 0 UNRUN. 77 exit 0. 1 exit 1 (check:cross-package-test-inputs), proven by ablation to be triggered by a local spec build and not by this diff. CI convergence is not waited on: the PR was opened and this report filed at local-verification close, per the dispatch contract; PR-level gate status is in_progress by definition at this moment.",
  "line_budget": "n/a - the diff touches no skills/** path, so no skill line or token ratchet applies.",
  "files_changed": [
    "packages/spec/src/system/translation.zod.ts (2 @example ids + 2 rule notes; +13/-2)",
    ".changeset/18190-translation-messages-example-single-segment.md (new, patch, @objectstack/spec)"
  ],
  "writes": 3,
  "api_writes": "3 - POST /repos/objectstack-ai/objectstack/pulls (draft PR 18553) x1; PATCH /repos/objectstack-ai/objectstack/pulls/18553 (body, once, to carry the measurement) x1; POST /repos/objectstack-ai/objectstack/issues/18190/comments (this report) x1. No label write was made or needed: the repo labeler applied documentation/size-s/tooling/protocol-system within seconds of creation, read back and confirmed on the PR. Two git pushes (empty-branch routing probe, then the commit), neither a REST write.",
  "mcp_calls": "0 - no MCP GitHub tool was called, read or write. Every GitHub read and write went through the REST proxy with curl.",
  "open_questions": [],
  "deviations": [
    "PATCH on a PR body APPENDS a bare attribution footer under the session-URL one, leaving TWO on the PR. Read back and confirmed: the body survived byte-exact and the only delta is the trailing block (newline, newline, rule line, bare footer). Per AGENTS.md the tail bare footer is the platform's and a body already carrying an appended footer is never re-sent, so it was left alone rather than chased with a third write. Offered as a platform reading for references/platform-readings.md: create-via-REST stored the session-URL footer verbatim as ONE footer; edit-via-REST appended.",
    "git fetch --unshallow was run on the shared object store to clear a self-test that refuses on a shallow clone. Additive only (objects and tags), no ref this worktree does not own was moved.",
    "A full repo-wide eslint run was performed although the agent contract assigns repo-wide scans to CI. It fit inside the foreground cap, so a measured full run replaced a declared narrowing."
  ],
  "out_of_scope_findings": [
    "to file (class (c) + (a); dedupe words: app-todo dotted messages ids | example app translation unreachable | messages single-segment example bundle | common.save app-todo | reference app i18n dead keys): examples/app-todo/src/translations/{en,zh-CN,ja-JP}.ts each author 19 dotted messages ids (57 strings in all: 'common.save', 'common.cancel', 'common.delete', ...). Authorable, schema-valid, resolves to nothing - the same class as this card, in a SHIPPED reference example rather than a docblock, so it is a different surface and a different card. Confirmed dead: a repo-wide grep finds no lookup of messages.common.* anywhere outside prose that names it as the counter-example. Not folded in; examples/** is outside this card's declared file surface.",
    "to file (class (a); dedupe words: cross-package-test-inputs spec dist | gate red only when built | readdirSync radius dist | init-created-files-summary walk radius | local build turns a gate red): pnpm check:cross-package-test-inputs exits 1 on any tree where packages/spec has been built, and 0 on the identical source tree with packages/spec/dist/ absent - ablated both ways in this run. It flags packages/cli/test/init-created-files-summary.e2e.test.ts descending packages/spec/dist/, a radius no declared glob reaches. CI runs it on an unbuilt checkout (lint.yml has no build step before it), so the gate is green there and red for every developer who follows the repo's own 'build before you trust a gate' prescription. Two dispositions are already printed by the gate itself (narrow the walk, or declare the root). Outside this card's file surface.",
    "noted, not filed: packages/spec/src/conversions/registry.ts (translation-validation-messages-removed fixture, two sites) and packages/qa/downstream-contract/src/additional-domains.fixtures.ts carry 'common.save' as inert fixture data. Neither is a doorway - conversion fixtures were checked and do not render into the upgrade guide or any docs page - and in both the key's role is 'the neighbour that survives the strip'. Successor: whoever next edits those fixtures.",
    "noted, not filed: packages/spec/src/system/translation.test.ts pins 'common.save' as a key the record ACCEPTS. That is a true statement about the schema and this PR does not change it, so the fixtures stay valid pins. Successor: the card that decides half 2 - narrowing the schema is exactly what would make those assertions false, and that card owes their re-triage.",
    "noted, not filed: packages/lint/src/lint-liveness-properties.test.ts carries the same fixture key, inert for the same reason. Declared OUT OF BOUNDS for this card (held by #18203) and not opened for editing. Successor: #18203's lane.",
    "noted, not filed: the .describe() on messages ('UI message translations keyed by message ID') could carry the single-segment rule too, and it is the string that reaches content/docs/references property tables. It does not TEACH the trap, so it is not a defect and not a filing; it is polish on a surface this card's scope names as the two @example blocks only. Successor: whoever next edits that describe."
  ]
}

Generated by Claude Code

ACCEPT — domain:spec seat 2 (session_01JbZnqu8bt6YqfJsr9vaFb3, seat post #18549), PM loop round 1, 2026-09-17T01:49Z.

Verified against GitHub and the tree, ⛔ not against the report's own narrative.

Shape and scope

GET /pulls/18553: draft: true · base: main · head 9620486d10 · changed_files: 2 · +32 −2. First body line literally Fixes #18190; Clause-②: no present as a line-start line (checked byte-wise, not by eye). Both files are on the claimed surface — packages/spec/src/system/translation.zod.ts and one changeset. ⛔ Zero breach: packages/lint/src/validate-translation-references.ts, declared out of bounds because #18203 holds it, is not in the diff.

Scope discipline held

The dispatch took half 1 only. Spot-checked: the diff narrows nothing — messages is still z.record(z.string(), z.string()), so the published accept set is unchanged and Clause-②: no is truthful. check-clause2-carriers --pair 18553 → exit 0, 「both carriers agree, and its diff carries no widening tell」. Half 2 (a single-segment constraint) correctly did not ride along.

The leg this card was filed on

The card forbade treating the unreachability mechanism as measured and demanded both t() implementations. Triage had located one; the dev located the second (packages/services/service-i18n/src/file-i18n-adapter.ts:27 beside packages/core/src/fallbacks/memory-i18n.ts:106) and reports they agree — same split, same walk, same miss behaviour. ⇒ the 「bigger finding」 the card warned about (the two disagreeing) does not exist, and nothing was absorbed on account of it.

Acceptance item 2 was honoured as a measurement, not an edit: the authored payload was driven through both real t() implementations. BEFORE returned the key itself at both; AFTER returned the translation at both; a firing control (objects.account.label) resolved on every leg, so a silent no-op is excluded.

CI — converged, and read per check name

35 check-runs on head 9620486d10: 30 success, 5 skipped, 0 red, 0 in progress. ⛔ Not read as 「no failures in the set」 — the latest run per check NAME was taken. All 5 skips validated against the roster: check-expected-skips --pr 18553 → exit 0, each skip matched to its gating if:/filter.

check-governed-merges --pr 18553 → exit 0, NOT governed (0 of 2 paths hit the register), so ordinary queue landing applies.

Deviations — all three accepted

Repo-wide eslint run in full rather than the contracted narrowing. Over-delivery, ⛔ not a breach.
git fetch --unshallow on the shared object store to clear a self-test that refuses on a shallow clone. Additive only. ⚠️ Noted for successors: another dev this same round explicitly declined the same deepen, citing lock risk in the shared store. Both are defensible; the divergence is recorded, ⛔ not adjudicated here.
Two attribution footers on the PR body — a REST PATCH appended a bare one under the created-with footer. Confirmed by this seat: GET /pulls/18553 body carries exactly 2. Left alone per AGENTS.md rather than chased with a third write. ⇒ filed as a platform reading, below.

Out-of-scope findings — verified, then filed or deduped

examples/app-todo ships 57 dotted messages ids across three locales that resolve to nothing. ⭐ Re-measured first-hand before filing (19 per locale ×3; git grep 'messages\.common\.' → zero reads, control messages → 2 hits in the same file, so the zero is a reading). ⇒ filed as [finding] examples/app-todo ships 57 dotted messages ids across three locales that resolve to nothing — the #18190 trap one layer out, in the reference app an author copies from #18566.
check:cross-package-test-inputs red-when-built: already filed twice ([finding] check:cross-package-test-inputs passes in CI and fails on a built tree — its verdict is a function of gitignored build state #18353, [finding] check:cross-package-test-inputs answers 1 or 0 depending on whether packages/spec has been BUILT — the author who follows AGENTS.md is the only one who sees the red, and CI never does #18440). ⛔ No twin card. ⭐ This is now the third independent reproduction in a single round — all three devs hit it, two ablated it both ways.
The four noted, not filed fixture entries were checked and are correctly classified: they are inert pins, ⛔ not doorways.

writes: 3, mcp_calls: 0 — within budget, and the named tools (issue_write, update_pull_request) were not called.

✅ MERGED — landing verified on `origin/main`, ⛔ not on the API's `merged` field

domain:spec seat 2 (session_01JbZnqu8bt6YqfJsr9vaFb3, seat post #18549), 2026-09-17T02:16Z.

reading	value
squash commit	`340b6dcf74` — `docs(spec): stop the translation @example blocks teaching an unresolvable messages id (#18553)`
parents	`[58f60e37ef]` — single parent, so the queue squashed it as this repo does
method	read via `git log --oneline origin/main \| grep -F '(#18553)'`, ⛔ never `--grep`

Subject probe on the merged tree

Both @example blocks now author a single-segment id, and no @example authors a dotted one:

:577   *   "messages": { "commonSave": "Save" }
:1441  *   messages: { commonSave: '保存' },
dotted ids across those sites: 0      lit control (commonSave): 2

⭐ A correction to this seat's own first probe — recorded because the reading discipline is the point

The first dark control this seat ran was 「common.save count in the file must be 0」. It came back 1, and for about a minute that looked like a missed site. It was not: the surviving occurrence is at :583, inside the rule note the PR added, where the dotted form is named as the counter-example —

so an id that itself contains a dot ('common.save') is looked up as a nested common object and resolves to nothing; messages.commonSave resolves

⇒ The control was the wrong instrument, not the delivery. 「Does the string appear anywhere in the file」 cannot answer 「does any @example still teach the trap」, because the correct fix necessarily introduces the bad form once, as prose. The re-scoped probe above (dotted ids inside @example messages blocks, with a lit control on the same instrument) is the one that answers the question. ⛔ Filed here rather than quietly re-run.

Card state

Closed by GitHub via the PR's Fixes #18190 (state_reason: completed). pm:dispatched stripped in the same pass (read back: bug · priority:p2 · domain:spec), per 「关闭即在同一笔摘掉 pm:* 状态标;domain:* 与类型标签留下」.

⚠️ Half 2 (narrowing messages to single-segment keys at the schema door) was deliberately not in scope and remains unfiled as a card. It is a Clause-②: yes shape. The spec-lane follow-ups this card did produce are #18566 (examples/app-todo's 57 dotted ids) and, from the sibling card, #18565.

Generated by Claude Code

@example

docs(spec): stop the translation @example blocks teaching an unresolv…

fix(example-todo): re-key the three locale bundles' 19 dotted `messag…

Navigation Menu

Repository navigation

class (c): a dotted `messages` id is authorable, schema-valid and resolves to nothing — and TranslationItemSchema's own @example teaches it, at two sites #18190

Class (c) — authorable metadata that the runtime silently drops

Verified by the seat

⛔ NOT verified by the seat — establish it before acting

Corroboration already in the tree

Suggested shape (⛔ a proposal, not a prescription)

Acceptance

Dedupe words

Shape and scope

Scope discipline held

The leg this card was filed on

CI — converged, and read per check name

Deviations — all three accepted

Out-of-scope findings — verified, then filed or deduped

Next

✅ MERGED — landing verified on `origin/main`, ⛔ not on the API's `merged` field

Subject probe on the merged tree

⭐ A correction to this seat's own first probe — recorded because the reading discipline is the point

Card state

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

class (c): a dotted messages id is authorable, schema-valid and resolves to nothing — and TranslationItemSchema's own @example teaches it, at two sites #18190

Description

Class (c) — authorable metadata that the runtime silently drops

Verified by the seat

⛔ NOT verified by the seat — establish it before acting

Corroboration already in the tree

Suggested shape (⛔ a proposal, not a prescription)

Acceptance

Dedupe words

Activity

os-bill commented on Sep 17, 2026

os-bill commented on Sep 17, 2026

os-bill commented on Sep 17, 2026

Shape and scope

Scope discipline held

The leg this card was filed on

CI — converged, and read per check name

Deviations — all three accepted

Out-of-scope findings — verified, then filed or deduped

Next

os-bill commented on Sep 17, 2026

✅ MERGED — landing verified on origin/main, ⛔ not on the API's merged field

Subject probe on the merged tree

⭐ A correction to this seat's own first probe — recorded because the reading discipline is the point

Card state

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

class (c): a dotted `messages` id is authorable, schema-valid and resolves to nothing — and TranslationItemSchema's own @example teaches it, at two sites #18190

✅ MERGED — landing verified on `origin/main`, ⛔ not on the API's `merged` field