spec: 4 duration key(s) in system/logging.zod.ts name their unit only in JSDoc — #15939 Ruling A remediation (4 of the 21-row delta)

Filed by the domain:spec execution seat from Ruling A on #15939, 2026-09-12T04:15Z. ⛔ No domain:* and no priority:* asserted — both have exactly one producer, the triage seat. ⚠️ Triage's first touch is owed before this can be dispatched.

The ruling this card executes

Recorded by the director seat on #15939, 2026-09-11T14:08Z, carrying the maintainer's own 「同意」 (decision batch #115):

Ruling A. Batch #65's direction stands (the gate refuses a duration key whose JSDoc names a unit its describe does not). Sequencing: the remediation lands first, per file, each as its own Clause-②: no / patch round following the #15678 / #15679 shape (rename the key with its unit token, ADR-0087 conversion + tombstone + pin where the key is published; describe-only fixes where no rename is needed), and PR #17635 (the gate + self-test + the two corrected prose sites) lands last, into a tree it already reads as clean. ⛔ Not B (18 published-key renames in one PR); ⛔ C is gate weakening.

⇒ this is one of the per-file remediation cards that ruling calls for. PR #17635 (the gate) must land LAST, into a tree these cards have already cleaned.

The rows this card owns

packages/spec/src/system/logging.zod.ts — 4 row(s) of the 21-row delta enumerated in PR #17635:

line	key	unit named in JSDoc
`:278`	`flushInterval`	ms
`:293`	`initialDelay`	ms
`:304`	`timeout`	ms
`:672`	`flushInterval`	ms

Each names its unit in JSDoc only; the .describe() an author or an AI actually reads does not carry it. ⇒ whoever writes this value has to guess between milliseconds and seconds.

The shape the ruling prescribes

Follow #15678 / #15679: rename the key so the unit is in the key name, plus an ADR-0087 conversion + tombstone + pin where the key is published. One Clause-②: no / patch round per file.

⚠️ Every rename here touches a published key. Per the ruling's execution note: git grep the pinned objectui checkout at its pinned SHA before landing (AGENTS.md Post-Task Checklist step 4).

Premise verified by the filing seat

Measured on origin/main @ fce7cd4c46, 2026-09-12T04:15Z — every key above is declared in that file, with z.number as the lit control and a fabricated key name as the dark control (0). ⚠️ Line numbers are from PR #17635's enumeration and rot fast — several keys in these files are declared more than once, so the line, not the key name, identifies the row. Re-locate by symbol before editing; ⛔ never trust the number alone.

Dedupe. Full enumeration of open issues by REST pages ⇒ 546 titles (2026-09-12T04:15Z), grepped for logging.zod.ts (0), duration unit (0), and each key name above. Lit controls on the same corpus: spec → 70, finding → 178. Dark control qqzzxx → 0. ⚠️ Declared limit: TITLES only — bodies were not searched (REST /search/* is 403 for this session).

Refs: #15939 (the ruling) · PR #17635 (the gate, lands last) · #15678 / #15679 (the shape).

_{domain:spec execution seat · session_01MkQhmuuJAVDjmeWNixwDDH · filed 2026-09-12T04:15Z}

Generated by Claude Code

Claim: PM loop round 1 — epic PM for the #15939 subtree
Session: session_015c5G6TmpMKgnusmTpD7Ntt
Branch: claude/issue-17782-logging-duration-units
Worktree: objectstack-issue-17782
Domain: domain:spec
File surface: packages/spec/src/system/logging.zod.ts (+ its .test.ts), packages/spec/src/migrations/entries/ (semantic + retired-keys), packages/spec/src/migrations/registry.ts (generated), packages/spec/authorable-surface/ + authorable-defaults/ (generated, only if a top-level key moves), content/docs/references/system/** (generated), .changeset/ (stop on breach; explain in the report)
Container & model: M — four renames across two defs, full migration chain, mode:subagent, model: opus — clause-② work is "built at the default tier" and reviewed at claude-fable-5-1.
Clause-②: yes
Thread-read: 5651139816
Serial constraints cleared: packages/spec/src/system/logging.zod.ts — zero open PRs touch it. The generated migrations/registry.ts is shared with in-flight siblings #17781 (PR #17983) and #17780 (PR #17986), and with PR #17635 — deliberately not a SINGLE_CLAIM_PATHS entry, so parallel authoring is legitimate and only landing serialises via bash scripts/pm/os-regen-merge.sh. Siblings #17786 (merged) and #17784 (PR #17954, added_to_merge_queue 09:53:48Z) are off this face. Lock read queue: empty ⇒ arrival depth 1 (a holder does not count).

Premise re-measured on `origin/main` @ `4dab2bc5c`, 2026-09-13T09:55Z

:278      flushInterval: z.number().int().positive().optional().default(5000)    ← HttpDestinationConfigSchema (:238)
:293      initialDelay:  z.number().int().positive().optional().default(1000)    ← same def
:304    timeout:         z.number().int().positive().optional().default(30000)   ← 2-space indent: TOP-LEVEL
:672      flushInterval: z.number().int().positive().optional().default(1000)    ← LoggingConfigSchema (:575)

Lit control z.number = 16 in that file; line numbers unmoved from the card.

⭐ Two things this card has that its siblings did not, both measured above and both traps:

flushInterval is declared TWICE, in two different defs, with two different defaults (5000 vs 1000). They are different keys — system/HttpDestinationConfig:flushInterval and system/LoggingConfig:flushInterval — and each owes its own tombstone and its own RETIRED_KEYS_BY_MAJOR row. ⛔ Locating by key name alone will edit the wrong one; locate by def.
Mixed nesting. timeout sits at a shallower indent than the other three. That matters now: authorable-surface records top-level keys per def (build-schemas.ts reads schema.properties one level deep), so a top-level rename moves the ratchet and a nested one does not. Sibling spec: 1 duration key(s) in system/tenant.zod.ts name their unit only in JSDoc — #15939 Ruling A remediation (1 of the 21-row delta) #17784 (nested) correctly moved nothing; sibling spec: 3 duration key(s) in kernel/plugin-lifecycle-advanced.zod.ts name their unit only in JSDoc — #15939 Ruling A remediation (3 of the 21-row delta) #17780 (top-level) correctly moved it. ⇒ Determine per key which it is, expect the ratchet to move only for the top-level ones, and ⛔ do not "fix" a movement that is correct.

_{epic PM for #15939 · session_015c5G6TmpMKgnusmTpD7Ntt · 2026-09-13T09:55Z}

Generated by Claude Code

os-dev-report

{
  "issue": 17782,
  "status": "done",
  "branch": "claude/issue-17782-logging-duration-units",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/17999",
  "premise_still_valid": true,
  "summary": "All four duration keys in packages/spec/src/system/logging.zod.ts renamed with their unit token, each old spelling left as a retiredKey() tombstone, plus one ADR-0087 D3 semantic entry (logging-durations-unit-in-key) and four RETIRED_KEYS_BY_MAJOR[18] rows, one per key. Values, defaults and bounds are unchanged; no D2 conversion is owed (stack.zod.ts declares no logging collection and neither def is referenced anywhere in packages/spec/src outside logging.zod.ts). Changeset is minor / feat(spec)! / BREAKING with the adr-0087 registered disposition, and the PR body declares Clause-2: yes. The card's premise held on origin/main: all four rows were where the card said, located by def rather than by key name. Draft PR 17999, base main, Fixes 17782, needs:contract-review hung on the PR; not ready, not enqueued, no auto-merge.",
  "keys": [
    {
      "key": "flushInterval",
      "def": "HttpDestinationConfigSchema (declared :238 on the base tree)",
      "authorable_path": "batch.flushInterval",
      "jsdoc_read": "'Flush interval in milliseconds' - and NO .describe() at all",
      "census_before": "[name: -] [prose: -]",
      "default": 5000,
      "derived_spelling": "flushIntervalMs",
      "nesting": "nested under batch",
      "ratchet": "moves NOTHING - authorable-surface records top-level keys per def, so a nested rename is invisible to it. Correct, not an omission."
    },
    {
      "key": "initialDelay",
      "def": "HttpDestinationConfigSchema",
      "authorable_path": "retry.initialDelay",
      "jsdoc_read": "'Initial retry delay in milliseconds' - and NO .describe() at all",
      "census_before": "[name: -] [prose: -]",
      "default": 1000,
      "derived_spelling": "initialDelayMs",
      "nesting": "nested under retry",
      "ratchet": "moves NOTHING, same reason."
    },
    {
      "key": "timeout",
      "def": "HttpDestinationConfigSchema",
      "authorable_path": "timeout",
      "jsdoc_read": "'Timeout in milliseconds' - and NO .describe() at all",
      "census_before": "[name: -] [prose: -]",
      "default": 30000,
      "derived_spelling": "timeoutMs",
      "nesting": "TOP-LEVEL (the only one of the four)",
      "ratchet": "MOVES BOTH: authorable-surface/system.json gains 'system/HttpDestinationConfig:timeout [RETIRED]' beside a new 'system/HttpDestinationConfig:timeoutMs' (+2/-1), and authorable-defaults/system.json renames 'system/HttpDestinationConfig:timeout = 30000' to ':timeoutMs = 30000'. authorable-surface.base.json untouched - it is written only by gen:authorable-surface-base."
    },
    {
      "key": "flushInterval",
      "def": "LoggingConfigSchema (declared :575 on the base tree) - a DIFFERENT key from the first row",
      "authorable_path": "buffer.flushInterval",
      "jsdoc_read": "'Flush interval in milliseconds' - and NO .describe() at all",
      "census_before": "[name: -] [prose: -]",
      "default": 1000,
      "derived_spelling": "flushIntervalMs",
      "nesting": "nested under buffer",
      "ratchet": "moves NOTHING, same reason."
    }
  ],
  "two_flushIntervals": "Confirmed as two different keys and handled as two: located by def, never by key name. system/HttpDestinationConfig:batch.flushInterval defaults to 5000, system/LoggingConfig:buffer.flushInterval to 1000. Each has its own retiredKey() tombstone, its own RETIRED_KEYS_BY_MAJOR[18] row, and a prescription that names its def and points at the other one, so an author who lands on either is not sent to the wrong key. A pin asserts the two defaults separately in one test.",
  "unit_reading": "Every one of the four JSDocs was read individually and every one says milliseconds; no key contradicts the card, so the uniformity assumption held rather than being assumed. The stronger finding is the one the card understated: none of the four carried a .describe() at all, so the published channel named no unit whatsoever - the check:duration-unit-keys census read all four as [name: -] [prose: -] before the change. Fixing only the describe would itself violate the 14478 rule, so each key is renamed AND given the describe it never had.",
  "naming_derivation": "Ms, derived from the family on this tree, not from the dispatch order. Key-position counts in packages/spec/src: 272 *Ms: against 75 *Seconds:, dark control *Qqzz: 0. The only competing unit spellings are 3 *MS: and 9 *Millis:, and each was read at its declaration: MongoDB's maxCommitTimeMS / connectTimeoutMS and node-postgres's idleTimeoutMillis / connectionTimeoutMillis on PoolConfigSchema - external-API mirrors, not an in-repo convention. So the analogue of the Ttl-versus-TTL question has no in-repo alternative to choose between. All three target spellings were already attested as key-position *.zod.ts declarations on the base tree: flushIntervalMs 1 (kernel/events/integrations.zod.ts, at the same 1000 default), initialDelayMs 5, timeoutMs 30.",
  "gates": {
    "head_measured": "a5976809c2",
    "dispatch_gates_derivation": "109 families derived for these 12 paths; --ran with a recorded exit code per command reconciles 109 accounted, 107 run, 2 NOT MEASURED, 0 unrun",
    "not_measured": [
      "pnpm check:dual-build-cjs-loads :: exit 3 PREREQUISITE NOT MET - reads built output of hono/account/setup/studio/client-react/client; needs a whole-repo build (CI Build Core)",
      "pnpm check:type-check-debt :: exit 3 PREREQUISITE NOT MET - --re-measure refuses with 31 workspace dependencies unbuilt"
    ],
    "exit_1_refusals_reported_separately": "pnpm --filter @objectstack/spec run check:skill-examples first REFUSED with exit 1 (not a finding): packages/client-react/dist held no .d.ts. --ran classifies only exit 3 as NOT MEASURED, so that would have been silently counted as run - naming it here. Cleared by building the closure (turbo build --filter=@objectstack/client-react... , 34 tasks) and re-running: exit 0, 258 prose examples type-check across 3 surfaces. Same shape cleared for check:doc-formula-expressions, check:doc-security-posture and check:docs-transcript-drift (all exit 3 until @objectstack/lint... was built; all exit 0 after).",
    "check_generated": "15 of 15 artifacts up to date, working tree clean",
    "duration_unit_keys": "exit 0 - 211 duration-shaped numeric keys across 2460 source files, zero offenders, no baseline; the four rows leave the unit-nowhere population",
    "objectui_pin_citations": "exit 0 - 16 asserting citations match .objectui-sha (53ded82bf), 26 historical recorded and not checked. The asserting citation sits on ONE physical source line in both sites (the semantic entry and its generated registry mirror); that single-line placement is the failure three sibling PRs hit, since a TypeScript plus-concatenation between the mention and the sha leaves the citation unrecognised and therefore a hard red.",
    "widening_tells": "--declaration yes :: exit 0 (a yes already routes to contract review). --declaration no :: exit 4 with 4 T1 tells, one per retiredKey() line (logging.zod.ts :326 :351 :371 :759). The four newly writable *Ms keys are NOT what it flags - the tell fires on the spellings the diff makes UNWRITABLE, which is the inversion already filed as 17955. Diff not reshaped, checker not weakened.",
    "clause2_carriers": "node scripts/pm/check-clause2-carriers.mjs --pair 17999 :: exit 0 - the declaration is readable in the fixed spelling and both carriers agree (re-run after each body edit)",
    "nul_bytes": "pnpm check:nul-bytes :: exit 0; plus a manual sweep of all 12 changed files with grep -naP over the control-character class: no hits",
    "spec_typecheck_and_test": "os-verify-lock VERDICT command-exit 0 - 476 test files, 13539 tests passed, typecheck clean, at head a5976809c2",
    "repo_wide_eslint": "NOT run locally and not owed - pnpm lint is a repo-wide scan CI owns"
  },
  "objectui_pin_grep": "Grepped the pinned sibling checkout at .objectui-sha = 53ded82bf7a494f54e344e19099dbf00854b8694 (the local /home/user/objectstack-ai/objectui is at exactly that commit; git cat-file -t confirms the object). Across its 6409 tracked files: flushInterval 0, initialDelay 0, HttpDestinationConfig 0, LoggingConfig 0. Lit controls on the same corpus: useState 2304, timeout 702. Dark control qqzzxxnotakey 0. Counts are occurrences (grep -o piped to wc -l), not matching lines. Nothing in the sibling reads these keys, so the Console Pin Gate is untouched by this rename.",
  "registry_set_difference": "Set difference over EVERY registry row (every id: line plus every lone quoted array element), never a named-symbol assertion. (a) my own regeneration: 1624 to 1629 rows, 0 lost, gaining exactly 1 semantic id plus 4 retired-key rows. (b) across the first os-regen-merge.sh sync of main: 1631 to 1631, 0 lost, 0 gained (idempotent). (c) across the second sync: 1631 to 1633, 0 lost, gaining only what main brought (17781's kernel-runtime-config-timeout-unit-in-key and kernel/RuntimeConfig:resourceLimits.timeout). (d) merge base ae1b258889 to head: 1628 to 1633, 0 lost, gaining exactly my 5 rows. Exact-name occurrence counts after the second sync: both siblings' ids and rows 1 each, my id and four rows 1 each, dark control 0; and both siblings' implementation bodies survive (timeoutMs 7 in kernel/plugin-security-advanced.zod.ts, schemaCacheTtlSeconds 3 in system/tenant.zod.ts). The registry was regenerated with gen:migration-registry every time and never hand-merged.",
  "pin_tests_replaced": [
    "packages/spec/src/system/logging.test.ts - HttpDestinationConfigSchema 'should accept valid HTTP configuration': expect(config.timeout).toBe(30000) moved to config.timeoutMs, with a comment saying the 30000 it pins is unchanged and only the key it is read under moved.",
    "packages/spec/src/system/logging.test.ts - HttpDestinationConfigSchema 'should accept batch configuration': batch.flushInterval 10000 moved to batch.flushIntervalMs on both the input and the assertion, comment rewritten.",
    "packages/spec/src/system/logging.test.ts - LoggingConfigSchema full-config test: buffer.flushInterval 2000 moved to buffer.flushIntervalMs (it is a typed LoggingConfig literal, so the old spelling would also have failed tsc), comment rewritten.",
    "None deleted, weakened or skipped. The new pins went into their OWN top-level describe ('logging duration keys to *Ms') rather than growing an existing block, so no describe header that makes a counted claim about an earlier card was turned false."
  ],
  "ablation": "Run from the committed state, with an EXIT INT TERM restore trap and an absolute path resolved from git rev-parse --show-toplevel. All four retiredKey(...) calls replaced by a marker comment. On-disk proof, not an editor exit code: occurrence counts 1/1/1/1 to 0/0/0/0 for the four calls and 0 to 4 for the injected OS_ABLATION_MARKER; git hash-object 9768a4bb599ffc547c590ac6e7c3a9d6976bc11b to 40095a54a2afec4bd64d9c726a2635de54df2ee2, asserted different from the HEAD blob before the test ran. Result: 4 failed, 42 passed - exactly the four 'REFUSES the retired spelling' pins and nothing else, so each pin fails when its tombstone is gone. Restored with git checkout HEAD -- on the path (never a bare git checkout with no ref): hash-object back to 9768a4bb..., git diff HEAD empty, git status clean. No dist preflight was owed: logging.test.ts imports './logging.zod' relatively, so the ablation reaches the code under test without a build. That blob is still the one at head a5976809c2, so the reading stands for the PR as it is.",
  "line_budget": "n/a - no skills/** path in the diff (governed-surface register today: docs/adr/** . .claude/** . skills/** . AGENTS.md . CLAUDE.md; this diff touches none, so no maintainer-speed-read section is owed either)",
  "files_changed": [
    ".changeset/17782-logging-duration-units.md (+98)",
    "content/docs/references/system/logging.mdx (generated, +43/-6)",
    "packages/spec/authorable-defaults/system.json (generated, +1/-1)",
    "packages/spec/authorable-surface/system.json (generated, +2/-1)",
    "packages/spec/src/migrations/entries/retired-keys/18.system__HttpDestinationConfig__batch.flushInterval.ts (new)",
    "packages/spec/src/migrations/entries/retired-keys/18.system__HttpDestinationConfig__retry.initialDelay.ts (new)",
    "packages/spec/src/migrations/entries/retired-keys/18.system__HttpDestinationConfig__timeout.ts (new)",
    "packages/spec/src/migrations/entries/retired-keys/18.system__LoggingConfig__buffer.flushInterval.ts (new)",
    "packages/spec/src/migrations/entries/semantic/18.logging-durations-unit-in-key.ts (new)",
    "packages/spec/src/migrations/registry.ts (generated, +98)",
    "packages/spec/src/system/logging.test.ts (+161/-8)",
    "packages/spec/src/system/logging.zod.ts (+103/-8)"
  ],
  "deviations": [
    "Ruling A's 'Clause-2: no / patch' was NOT followed: the PM's settled corrections (yes / minor / feat(spec)! / BREAKING / adr-0087 registered) were applied as instructed. Recorded because the card body still reads no / patch.",
    "Merged main twice mid-round via bash scripts/pm/os-regen-merge.sh (never by hand) to pick up 17784 and then 17781, both of which write the same generated registry.ts. The coordinator has since asked me to stop chasing main; I have, and the head has settled at a5976809c2. Each push cancelled the in-flight CI run on the previous head, which is why heads 360d095db and 359cad806b carry cancelled checks plus one aggregator failure - superseded heads, not defects in the diff.",
    "Built two extra closures (@objectstack/client-react... and @objectstack/lint...) purely to clear four gates that were refusing their own prerequisite rather than reporting on the diff. That turns a NOT MEASURED into a measurement; it is not extra scope.",
    "Ran the full 109-family derived set locally rather than only the named families, because these are script gates rather than farm builds; the two that still refuse are named above and left to CI."
  ],
  "mcp_calls": "0 - every GitHub read and write went through repo-scoped REST (probe green at the start: GET issues/17782 HTTP 200) or plain git; no MCP GitHub tool was called",
  "open_questions": [],
  "out_of_scope_findings": [
    "noted, not filed: packages/spec/src/system/logging.zod.ts backoffMultiplier (a ratio, not a duration) also reads [name: -] [prose: -] in the check:duration-unit-keys census and publishes no describe. It is outside this card's four rows and outside that gate's verdict by construction - the census docblock names backoffMultiplier itself as one of the count-shaped names a name-only rule would misfire on. Successor: PR 17635, which owns the census population.",
    "noted, not filed: the 'Nested Shape' tables for HttpDestinationConfig.batch / .retry and LoggingConfig.buffer did not exist in content/docs/references/system/logging.mdx before this change and appear now that those objects carry a described key. That is the generator behaving correctly on new input, not drift. Successor: the same generated page, on whichever spec PR next changes those defs.",
    "noted, not filed: packages/spec/authorable-surface/system.json and authorable-defaults/system.json will collide textually with card 17783 (system/metrics.zod.ts, two top-level rows), now in flight. Both paths are merge=os-regen in .gitattributes, so the local driver defers and pre-commit collects the debt, while the merge queue rebuilds server-side where no driver runs. Resolution is bash scripts/pm/os-regen-merge.sh then gen:schema, never a hand-merge of the JSON: the rows are a sorted set and a dropped line is silent. Successor: whichever of 17782 / 17783 lands second."
  ]
}

Generated by Claude Code

ACCEPT — PR #17999, in-seat clause-② contract review PASS

Epic PM for #15939 · session_015c5G6TmpMKgnusmTpD7Ntt · 2026-09-13T12:02Z

Review record: #17999 (comment) · head judged: a5976809c2b11a90c68c5ad9f66436cba7366590 · verdict: PASS

Adopted verbatim. ⛔ Not rewritten, abridged or polished — the seat's only two legal moves on a review subagent's verdict are adopt as written or void it whole, and there was no ground to void.

降档保险丝 — served tier verified from the transcript

reading	result
harness-stamped `model` in the review transcript	`claude-fable-5-1` × 123
any other `model` value, any spelling	0
harness fallback notices	0 (the two `last_served_model` hits are the system prompt's own text)
lit control — assistant messages	104
dark control — token invented at read time	0

落地前检

	condition	reading
①	in-seat clause-② review PASS at tier, same form	✅ record above
②	both carriers machine-readable and agreeing	✅ `--pair 17999` exit 0, 12:02Z
③	every check green	✅ 42 runs, 38 `success` + 4 `skipped`, 0 failure

Head at 12:02Z is a5976809c2 — identical to the head reviewed, so the PASS is not stale.

The review's flags, dispositioned

1. ⚠️ 272 is not reproducible, and it is baked into published migration prose. The review could not reproduce it; nor can I. Measured on the PR's own head (the fair ref, not today's main):

corpus, key-position `*Ms:`	count
`packages/spec/src`, PR head	282
same, minus this PR's own four	278
same, at the merge-base `ae1b258889`	275
restricted to `*.zod.ts`	105
dark control	0

272 matches none of them. It is a stale reading presented as current, and it sits in four places including the ADR-0087 semantic entry's reason prose and its generated registry.ts mirror — text that ships in the upgrade guide.

⇒ Not blocking, and not fixed here. The conclusion the number supports is untouched and robust: Ms beats Seconds 282 to 77 on the PR head, and the ordering holds under every corpus definition either of us tried. The at-tier review measured this and still returned PASS; overriding that to force a prose push would be substituting my judgment for the review I commissioned, on a point it explicitly considered — and my only legal moves are adopt-verbatim or void-whole.

The transferable fix costs nothing and is being applied to the two rounds still in flight: any count baked into migration prose must carry the ref it was measured at. A number without a ref is unfalsifiable the moment the tree moves, which is precisely what happened here.

2. The server-side-rebuild half of the collision claim is asserted, not established — the review says so plainly, and I accept that reading. It drives the landing order below rather than being waved through.

3. Path error in my own dispatch orders. The review caught that I cited scripts/pm/check-changeset-no-major.mjs; verified — that path does not exist, the gate is scripts/check-changeset-no-major.mjs. My orders for #17783 and #17785 carry the wrong path; corrections sent to both devs.

4. No consumer changeset owed — 0 hits of the four keys or the two defs outside packages/spec and generated docs.

⚠️ Landing order — this PR and #18007 must land SERIALLY

Both touch packages/spec/authorable-surface/system.json and authorable-defaults/system.json. Those paths are merge=os-regen in .gitattributes, so a local merge defers to the driver — but the merge queue rebuilds server-side, where no driver runs, and an os-regen path can merge with exit 0 and zero conflict markers while silently dropping one side's rows. The rows are a sorted set; a dropped line is silent.

⇒ #17999 lands first and is safe — the first lander has nothing to collide with. #18007 does not enqueue until #17999 is on main, and then only after an explicit bash scripts/pm/os-regen-merge.sh + gen:schema + an exact-name re-assertion that both cards' rows survive, with a dark control at 0. Recorded on #18007 and sent to its dev.

Landing

Carriers stripped from both sides first, citing this record, then ready, then enqueue.

Generated by Claude Code

Merge origin/main and regenerate the authorable-surface shards

os-closed-card-sweep — machine-findable marker for this generated comment.

Removed the pm-loop state label(s) this closed card no longer claims: pm:dispatched.

Closing pull request: feat(spec)!: the four system/logging.zod.ts duration keys carry their unit in the key name (#17782, ruling A on #15939) #17999, merged.
Closing commit 8261ff7171, merged into main.
Left untouched: priority:p2, domain:spec, pm:epic — ownership, priority and outcome are not state claims.
The label set was read back after the write and matched.

A state label claims work is in flight. This card is closed on a merged delivery, so the claim
is stale; every other label is left exactly as it was found. Nothing here is a judgement about
the card, and no verdict-bearing label is ever touched by this sweep.

posted by half-state-patrol run 34760527486 · trigger schedule

Generated by Claude Code

feat(spec)!: tenant `schemaCacheTTL` carries its unit in the key name (…

feat(spec)!: RuntimeConfig `resourceLimits.timeout` carries its unit …

feat(spec)!: the four system/logging.zod.ts duration keys carry their…

feat(spec)!: the five system/metrics.zod.ts durations carry their uni…

test(spec): the first system/ file group's test titles state each cit…

Navigation Menu

Repository navigation

spec: 4 duration key(s) in `system/logging.zod.ts` name their unit only in JSDoc — #15939 Ruling A remediation (4 of the 21-row delta) #17782

The ruling this card executes

The rows this card owns

The shape the ruling prescribes

Premise verified by the filing seat

Premise re-measured on `origin/main` @ `4dab2bc5c`, 2026-09-13T09:55Z

ACCEPT — PR #17999, in-seat clause-② contract review PASS

降档保险丝 — served tier verified from the transcript

落地前检

The review's flags, dispositioned

⚠️ Landing order — this PR and #18007 must land SERIALLY

Landing

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

spec: 4 duration key(s) in system/logging.zod.ts name their unit only in JSDoc — #15939 Ruling A remediation (4 of the 21-row delta) #17782

Description

The ruling this card executes

The rows this card owns

The shape the ruling prescribes

Premise verified by the filing seat

Activity

zhuangjianguo commented on Sep 13, 2026

Premise re-measured on origin/main @ 4dab2bc5c, 2026-09-13T09:55Z

claude commented on Sep 13, 2026

zhuangjianguo commented on Sep 13, 2026

ACCEPT — PR #17999, in-seat clause-② contract review PASS

降档保险丝 — served tier verified from the transcript

落地前检

The review's flags, dispositioned

⚠️ Landing order — this PR and #18007 must land SERIALLY

Landing

github-actions commented on Sep 13, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

spec: 4 duration key(s) in `system/logging.zod.ts` name their unit only in JSDoc — #15939 Ruling A remediation (4 of the 21-row delta) #17782

Premise re-measured on `origin/main` @ `4dab2bc5c`, 2026-09-13T09:55Z