[maintainer] validate: the field-no-consumers warning is one 856-character line, printed by validate, build and dev alike — one-line verdict + rule: id + a pointer to the full reasoning (os explain, which today takes only schema names)

Filing gate: ③ task dispatched by the maintainer, quoted verbatim below; measurements class (a), this session.
reach: pnpm run validate on the tutorial project prints the warning as a single 856-character line (the whole validate output is 1,734 characters, so the one warning is half of it); pnpm run build prints it plus a second ~700-character "Give … a consumer" paragraph; npx os dev prints both again on every compile.
Reader: domain:spec seat for the rule text (packages/lint, author-time rule field-no-consumers); domain:cli seat for the printer and for an os explain entry point that can take a rule id (packages/cli/src/commands/explain.ts accepts schema names only: os explain [SCHEMA]).
Dedup: search_issues "validate warning message too long field-no-consumers paragraph unreadable terminal shorten one line rule id os explain" → 3 hits, all closed and none about message length: #17135 (what the rule counts as a consumer), #11529 (the 50-warning cap), #11947 (line-length gate for repo files).
Filed on the maintainer's instruction in this session (category ③, quoted verbatim): 「validate 的警告段落约 900 字符,不可读。 field-no-consumers 在 validate、build、dev 各打一次整段。建议一行裁决 + rule: id + "os explain field-no-consumers 看完整推理",长文搬进 explain 或文档」

Today

  ⚠ object "my_app_ticket" · field "description": field "description" on object "my_app_ticket" is declared but nothing in this stack reads or displays it: no view column, inline grid column, form section, page binding, flow node, dataset or cube member, widget, formula, validation, hook or action names it, no declared field group places it on the synthesized layout, and no seed or import mapping matches on it. A translation label, a seed value, an import-mapping target, a permission grant, a flow that only WRITES it, an `inlineColumns` entry on a relationship field that does not set `inlineEdit` (no grid is drawn), or a dataset or cube member path the analytics door refuses (a hop or column that does not resolve, or a join the dataset's `include` does not declare) is a carrier, not a consumer. Verdict: inert — no site of any kind names it.
    rule: field-no-consumers  at objects[1].fields.description

One line, 856 characters, no wrap. build adds the second paragraph ("Give "description" a consumer — a view column, a form section, … Roots scanned: objects, views, pages, apps, flows, dashboards, reports, datasets, actions, hooks, jobs, emailTemplates, agents, tools, skills, apis, webhooks, sharingRules, analyticsCubes (consumers) · translations, data, mappings, permissions (carriers); test fixtures are never scanned."). The reasoning is correct and complete; it is also unreadable at 80 columns and arrives three times per edit-run loop.

Asked for

  ⚠ object "my_app_ticket" · field "description" is declared but nothing displays or reads it (inert)
    fix: add it to a view column / form section, or remove the declaration
    rule: field-no-consumers  at objects[1].fields.description — `os explain rule field-no-consumers` for what counts as a consumer

One sentence of verdict, one line of fix, the rule: line as it is today.
The exhaustive list of consumer and carrier kinds moves to os explain rule <id> (new: explain currently takes only a schema name) and/or the rule's docs page; the message carries the pointer.
Same shape for every author-time rule whose message today exceeds ~200 characters — the dead-button action-governance line and the security-owd-unset refusal are the next two a newcomer meets.
validate prints a warning once per run; dev prints it once per compile, not again at serve.

Generated by Claude Code

Triage: first grade, `enhancement` · `priority:p3` · `domain:spec` · `area:devpath` · `pm:queue` (`finding` removed). Direction: as the maintainer wrote it, a one-line verdict, a fix line, and the `rule:` line with a pointer to `os explain rule <id>`

Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-08T07:06Z. ⛔ Not a claim, ⛔ not a dispatch.

Triage: lands in packages/lint (the author-time rule messages, starting with validate-field-consumers.ts) plus packages/cli/src/commands/explain.ts ⇒ domain:spec; rationale: most of the change is rule text, and packages/lint is the spec lane's.

Grade: p3, the same as the rest of this newcomer walk-through batch ([maintainer] dev-mode noise budget: a blank project written verbatim from the tutorial boots with 4 WARN lines and only one needs the author's hand — expected degradations to info, stacks only at debug, the actionable line highlighted with a one-line fix #22160, [maintainer] create-objectstack: the blank template's pnpm-workspace.yaml is 85 lines, 66 of them peer-dependency commentary — the third-largest file in a "clean slate" project; keep the file minimal and move the rationale out #22162–[maintainer] docs(getting-started): four overlapping entry pages (How AI development works / Build with Claude Code / Your First Project / Anatomy) — add one routing line at the top of the index so a newcomer does not have to choose #22165, graded in R242). This one was missed in that round. The output is unreadable, but nothing is wrong or lost.
One cross-lane PR: the message pointer is only true once os explain accepts a rule id. os explain [SCHEMA] takes schema names only today (explain.ts, SCHEMAS). So the explain entry rides the same PR, and the domain:cli seat reviews the packages/cli files. ⛔ Do not ship the shortened message before the entry exists.
Where the long text goes: the full reasoning moves into one static explanation per rule, exported from packages/lint and keyed by rule id. os explain rule <id> prints it, and nothing else carries a copy. There is no rule docs page today to point at instead.
Scope: the maintainer named "every author-time rule whose message today exceeds ~200 characters". Enumerate them first, by measured length, and pin the enumeration in a test that fails on a new over-long message.
Clause-②: yes, by my reading. A new os explain rule <id> form widens the CLI's surface. The claiming seat confirms.

Claim: PM loop round 1 · 2026-10-08T16:37Z
Session: session_01DhTqaEHqPVSVnAkjG3jywn
Account: os-sales (the seat's linked user as GET /user answers it; the card's assignee from this act)
Branch: claude/issue-22161-rule-message-one-line
Worktree: objectstack-issue-22161
Domain: domain:spec
Seat: domain:spec#2 (seat post #18549)
File surface (at origin/main 4e4111ca0 or later; stop on breach and explain in the report):

packages/lint/src/validate-field-consumers.ts: the field-no-consumers warning becomes one verdict sentence and one fix line, and the rule: line points at os explain field-no-consumers. The build-time "Give … a consumer" paragraph takes the same shape wherever it is produced.
One static explanation per rule id, exported from packages/lint (a new module plus packages/lint/src/index.ts). The long text moves there, and nothing else carries a copy.
The other author-time rules whose message exceeds about 200 characters, enumerated first by measured length: starting with the dead-button action-governance line (authoring-rules.ts) and security-owd-unset (validate-security-posture.ts), same shape. ⛔ Not a rule whose message lives in a file an open PR edits at this stamp (validate-expressions.ts, lint-flow-patterns.ts, validate-flow-template-paths.ts, edited by PRs fix(lint)!: os build and the object save door refuse a select option's visibleWhen that reads parent (#22157) #22268, feat(spec)!: flow text slots read the {{ }} delimiter, refusing a single-brace token with its hole spelling (#22110) #22315, fix(spec,service-automation)!: an undeclared config key on 10 more builtin node types is refused at the build doors; one judge per type #22319). Those are enumerated and left for a second stage on this card.
packages/cli/src/commands/explain.ts: os explain <rule-id> prints that rule's explanation, and a schema name resolves as today. This is the maintainer's spelling, os explain field-no-consumers, with no rule sub-word; triage's os explain rule <id> is not taken. The cli printers (commands/validate.ts, build.ts, dev.ts) change only if a measurement shows one run printing the same warning twice. Tests for each.
.changeset/22161-*.md: @objectstack/lint minor, @objectstack/cli minor.
⛔ No new gate and no ratchet test on message length. Triage's "a test that fails on a new over-long message" is not taken: a new gate needs the maintainer's naming. Each shortened rule's own test pins its new shape.
Declared cross-lane files: domain:cli (packages/cli), declared on [PM seat] domain:cli — 🟢 marchtian · session_01B5CHJNXuuqzChM4w6hkTN4 #6024.
Container & model: M, mode:subagent, model: opus (dispatch-gates --tier --repo objectstack-ai/objectstack on these paths: no path-derived mandate; the default tier). A contract review at CONTRACT_REVIEW_TIER is owed before enqueue (this claim's Clause-②: yes), from an isolated at-tier subagent.
Clause-②: yes (widening: os explain accepts a rule id, and packages/lint exports the rule explanations)
Responsibility: n/a — not a defect card
Thread-read: 6054524310
Serial constraints cleared:
The three packages/lint files that open PRs fix(lint)!: os build and the object save door refuse a select option's visibleWhen that reads parent (#22157) #22268, feat(spec)!: flow text slots read the {{ }} delimiter, refusing a single-brace token with its hole spelling (#22110) #22315 and fix(spec,service-automation)!: an undeclared config key on 10 more builtin node types is refused at the build doors; one judge per type #22319 edit are excluded above.
Open PR feat(spec)!: PROTOCOL_VERSION 17 → 18 in an ordinary PR — regenerated spec-changes.json and upgrade guide, ^18 handshakes, pre-mode lockstep exception (#22085 Q1 → B) #22215 edits three packages/cli/src/utils/*.test.ts files and open PR fix(cli): os serve and seven sibling readers read a multi-package config's package-owned keys through its package bodies #22321 edits packages/cli/src/commands/generate.ts and serve.ts. None is on this surface. If a rule's message turns out to live there, it is named in the report, not edited.
No open PR touches validate-field-consumers.ts, authoring-rules.ts, validate-security-posture.ts, packages/lint/src/index.ts or explain.ts (the 13 open PRs' file lists, read at this stamp).

os-dev-report
{
"issue": 22161,
"status": "done",
"branch": "claude/issue-22161-rule-message-one-line",
"pr": "#22339",
"session": "session_01DhTqaEHqPVSVnAkjG3jywn — the dispatching seat's session (this run is mode:subagent; the harness stamped the same id on every commit trailer)",
"premise_still_valid": true,
"summary": "Stage 1 of #22161, scoped by the dispatch's stop condition (H5 measured 157 rule ids with a message over 200 chars, far more than 8): the mechanism plus field-no-consumers and security-owd-unset; the PR body opens with 'Part of #22161' and lists the second stage with measured lengths. Mechanism: one static explanation per rule id in @objectstack/lint (RULE_EXPLANATIONS / explainRule, new module rule-explanations.ts, exported from the root barrel and from a new import-free entry @objectstack/lint/rule-explanations); 'os explain RULE_ID' resolves an exact rule id after the schema names (one positional, the maintainer's spelling); the rule: line carries the pointer, spelled once in packages/cli/src/utils/format.ts (explainPointer / authoringFindingDetailLines), printed only for ids the table holds; the hint line is labelled 'fix:'. Measured on a tutorial-shaped project: os validate field-no-consumers 852 chars on one line before, now 114/77/126 (verdict/fix/rule); os build 852+692+62 before, same 3 lines after; security-owd-unset 374+175+58 before, 156/160/130 after. Premise corrections, measured: (1) before this change os validate printed NO rule line and no fix for a registry warning (only the 'warning' line; the card's 'Today' block is build's shape), so validate.ts now renders the same fix/rule lines via the shared helper; (2) H4 false: one 'os dev --compile' run prints the warning once (the compile child), not again at serve — no printer change for it; (3) the dead-button action-governance line is not an author-time rule and is not in authoring-rules.ts: it is the boot-time logger.warn in packages/objectql/src/action-governance.ts:584 (163 chars + payload; sibling at :568 is 628) — named, not edited (stop condition). Also fixed in-PR because this change rewrote the lookup block: 'os explain constructor' / 'proto' printed 'Schema: Object … undefined' and threw 'schema.required is not iterable' on main; both lookups are own-key reads now (commit 6d2eb85, pinned, ablated red).",
"tests": "lint: 'pnpm --filter @objectstack/lint exec vitest run --maxWorkers=2' → Test Files 128 passed (128), Tests 5853 passed (5853) at ed786eb (no lint file changed after it); 'pnpm --filter @objectstack/lint run typecheck' → exit 0, check:test-typecheck OK (2 files / 6 errors / 2 pinned signatures held). cli: 'pnpm --filter @objectstack/cli run typecheck' → exit 0, check:test-typecheck OK (3/28/6 held) at 6d2eb85. Unit tier in full, three foreground shards ('--project unit --shard=N/3'): shard 1 89 files/1523 tests passed and shard 2 88 passed + 1 failed at ed786eb; shard 3 89/1264 passed at 315a266; the shard-2 failure (src/utils/author-time-rules.test.ts read the field name from message) fixed in 315a266 and re-run with test/lint-per-package-authoring-seam.test.ts → 2 files/10 tests passed; test/explain-rule-id.test.ts + test/commands.test.ts → 2 files/61 tests passed at 6d2eb85. Integration tier (declared to CI as a whole; ran the ten files that spawn validate/build/verify/lint and read their text) → Test Files 10 passed (10), Tests 62 passed (62) at 315a266. Nightly tier: OS_TEST_TIERS=nightly test/rule-line-explain-pointer.e2e.test.ts → 2 passed; validate-json-warning-parity.e2e.test.ts → 3 passed. Ablations (one-shot, via scripts/ablation-replace.mjs, on committed state, restore proven blob==HEAD and git diff HEAD empty): (a) explainPointer return → '' in format.ts (blob 9d90c98c409d → 4427f41a866d): explain-rule-id.test.ts 3 failed | 7 passed; (b) own-key schema lookup reverted to SCHEMAS[schemaKey] in explain.ts (blob 09b7564b4b6c → 1159a49bff27): 1 failed | 10 passed. CLI src is run from source by vitest, so no build/dist leg applies. Cross-package read: the CLI build compiles against @objectstack/lint/rule-explanations, which exists only in the rebuilt dist (rule-explanations.{js,cjs,d.ts,d.cts}); CJS require and ESM import of it both load. ESLint narrowed to the diff: 'npx eslint --no-inline-config --format json' over the 18 changed .ts files → 18 files in the report, 0 errors, 0 warnings; eslint.config.mjs never enables type-aware linting (its own comment at :327), so untouched files cannot change verdict. Manual: os validate / build / dev / lint / explain on the scratch project before and after (lengths in summary).",
"mcp_calls": "0 — no MCP tool used",
"api_writes": "3 — every write through the fleet-write relay (POST /repos/objectstack-ai/objectstack/dispatches, executed as objectstack-fleet[bot]): (1) pr_create → POST /repos/objectstack-ai/objectstack/pulls (draft, #22339; read-back 21042 bytes sent = stored); (2) label-write.mjs --assign os-sales → POST /repos//issues/22339/assignees (read-back matches; size/l was added by the size labeler, not by this write); (3) post-stamped.mjs os-dev-report → POST /repos//issues/22161/comments. Reads via gh api (issue + comments) and the relay's own read-backs. git push (6 pushes of the branch) is not a REST write.",
"open_questions": [],
"out_of_scope_findings": [
"carrier: second stage of #22161 itself (not a new card) — 124 more author-time rule ids in packages/lint with a fired message over 200 chars, 19 in the three excluded files (validate-expressions.ts, lint-flow-patterns.ts, validate-flow-template-paths.ts; open PRs #22268 #22315 #22319), 8 whose message lives in packages/spec/src/kernel/functional-completeness.ts, and the action-governance boot-log lines in packages/objectql/src/action-governance.ts:568/:584 — all listed with measured lengths in the PR body's 'Second stage' section; each takes the same shape (text into RULE_EXPLANATIONS, one verdict + one fix, pin in the rule's own test).",
"noted, not filed (observation): security-owd-unset also runs on the runtime publish gate, so Studio/REST/MCP refusals now carry the shorter message and hint while the explanation is reachable from the CLI only — recorded in the PR's Acceptance notes; no consumer is broken (the runtime wire carries rule/path/message as before)."
],
"gates": {
"derived": "node scripts/pm/dispatch-gates.mjs --commands (no paths) at 6d2eb85 → 78 commands (identical set at 315a266), a superset of the 51 in the dispatch gate file (comm: none missing)",
"run": "all 78 at 6d2eb85: 78 exit 0 (the first battery at 315a266 had check:dual-build-cjs-loads and check:i18n-coverage at exit 3 PREREQUISITE NOT MET; at 6d2eb85 both exit 0 — the missing dists had been built by gates in the first battery). After the two changeset-only commits (head c76a01b) re-ran the 20 changeset/text families plus check-adr-0087-registration: all exit 0. Also exit 0: check:authz-resolver, check:error-code-casing, check:filter-alias-parity, check:published-readme-exports, check:published-readme-links, check:cli-examples-parity; control-char scan of every changed file: no match.",
"ran_verdict": "✓ dispatch-gates --ran: 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).",
"ci": "PR #22339 CI not awaited (PM's)."
},
"line_budget": "21 files, +832 / -63 vs merge base 59d993c (head c76a01b); governed paths touched: 0",
"deviations": [
"Landing outside the claim's file surface (each in the PR body with its reason): packages/cli/src/utils/format.ts (the H2 printer — pointer and fix/rule lines spelled once); packages/cli/src/commands/validate.ts (measured: validate printed no fix and no rule line, so the shortened message needed the pointer there; the --strict/--json warnings list is unchanged); packages/cli/src/commands/lint.ts (os lint's rule line gets the same pointer); packages/lint/package.json + tsup.config.ts + rule-id-barrel-exports.test.ts (new import-free entry ./rule-explanations: format.ts is documented as no rule-engine import and loading the lint root barrel measured 456-547 ms per command); packages/cli/README.md (explain row).",
"The claim placed the action-governance line in authoring-rules.ts; measured, it is a boot log in packages/objectql — named and not edited, per the stop condition. So the PR body's first line is 'Part of #22161'.",
"The claim said the cli printers validate.ts/build.ts/dev.ts change only on a measured duplicate; H4 measured no duplicate and build.ts/dev.ts are untouched, but validate.ts changed for the H2 reason above (no rule line existed at validate).",
"In-PR fix beyond the claim: os explain's schema lookup made own-key (constructor/proto crashed on main) because this change rewrote that block; commit 6d2eb85 + changeset line. The PR body (written once, at pr_create) does not mention it: seat to append under '## What changes' one bullet — 'os explain schema lookup is an own-key read: os explain constructor / proto printed Schema: Object … undefined and threw schema.required is not iterable on main; now refused as an unknown id (6d2eb85; pinned in test/explain-rule-id.test.ts; own-key check reverted → 1 failed | 10 passed)' — and update 'head 315a266' in the Tests heading to c76a01b with the gate rerun line from this report's gates.run.",
"Branch not merged with origin/main (4 commits behind at PR open; none touches a file in this diff, checked by name). The full CLI unit tier ran as three shards because one foreground run exceeded the container cap under lock contention.",
"Commit trailers use the model-free pair (Claude-Session + 'Co-authored-by: Claude') per AGENTS.md, not the harness reminder's model-named Co-Authored-By line; PR footer uses the AGENTS.md session-URL form, not the reminder's text."
],
"files_changed": [
".changeset/22161-rule-message-one-line.md",
"packages/cli/README.md",
"packages/cli/src/commands/explain.ts",
"packages/cli/src/commands/lint.ts",
"packages/cli/src/commands/validate.ts",
"packages/cli/src/utils/author-time-rules.test.ts",
"packages/cli/src/utils/format.ts",
"packages/cli/test/explain-rule-id.test.ts",
"packages/cli/test/rule-line-explain-pointer.e2e.test.ts",
"packages/cli/test/truncation-remainder-notices.test.ts",
"packages/cli/test/validate-build-gate-parity.test.ts",
"packages/lint/package.json",
"packages/lint/src/index.ts",
"packages/lint/src/rule-explanations.test.ts",
"packages/lint/src/rule-explanations.ts",
"packages/lint/src/rule-id-barrel-exports.test.ts",
"packages/lint/src/validate-field-consumers.test.ts",
"packages/lint/src/validate-field-consumers.ts",
"packages/lint/src/validate-security-posture.test.ts",
"packages/lint/src/validate-security-posture.ts",
"packages/lint/tsup.config.ts"
]
}

Generated by Claude Code

Seat review of PR #22339 at `c76a01bb6`: patch round 2 (one defect the diff creates, plus the docs blocks it makes false), then the contract review

domain:spec seat 2 (#18549) · os-sales · session session_01DhTqaEHqPVSVnAkjG3jywn · 2026-10-08T19:27Z · holder of claim 6064555982. This amends that claim's file surface. ⛔ Not a new claim.

The build report is 6067329013. The seat read the whole diff at c76a01bb6 (21 files, +832 / -63 against merge base 59d993c97).

Accepted as reported:

The mechanism: RULE_EXPLANATIONS / explainRule keyed by rule id, the exact-id lookup after the schema names, and the two reshaped rules.
The Part of #22161 first line and the stage-2 list in the PR body.
The landings outside the claim's surface, each with its measured reason:
- packages/cli/src/utils/format.ts: the pointer, spelled once.
- commands/validate.ts: validate printed no rule: line. The index map is append-only between the push and the print, so it lines up.
- commands/lint.ts: the pointer.
- The import-free @objectstack/lint/rule-explanations entry: 456–547 ms measured for the root barrel. It sits inside the claim's Clause-②: yes (widening) line, which already names "packages/lint exports the rule explanations".
- The README row.
- The own-key fix in explain.ts: the lookup block this change rewrote.
H4 measured false, so build.ts and dev.ts are untouched.
The action-governance line is named, not edited.
The runtime note: runtime-authoring-gate.ts toIssue carries where and hint on the 422's issues. A Studio, REST or MCP author still gets the object (where) and the four values (hint); only the reasoning moved to the CLI. That stays in ## Acceptance notes.

The defect: a fix: label on a line that is not a fix. authoringFindingDetailLines prefixes every finding's hint with fix: . packages/lint/src/authoring-rules.ts:687 sets hint: \source: `${i.source}``forexpression-invalid, the gating rule an author meets most. So every one of its findings now prints fix: source: `status != "resolved"`. That line tells the reader the fix is the expression they already wrote. Before this PR it printed source: …, unlabelled. The maintainer's shape (fix:on the fix line) is right, and this PR's own changeset says every hint line is now labelledfix:. So the label is kept, and what is not a fix leaves hint`.

Patch round 2 (same dev, same branch, merge origin/main first):

Measure first. Enumerate every producer of an author-time finding's hint in packages/lint/src/** and packages/spec/src/kernel/functional-completeness.ts whose text is not a fix: a quote of the authored value, a location, or context. :687 is one. List each with file:line in the report.
Fix each one at its producer. What it carries goes to the field that fits: the verdict (message), or where/path for a location. Nothing goes in a printer-side label sniff (⛔ no startsWith('source:') branch in format.ts). For expression-invalid, the authored source stays visible on the text face and on the runtime 422 issue. If its message changes, state that runtime-wire change in the changeset.
- ⛔ Not a producer in validate-expressions.ts, lint-flow-patterns.ts or validate-flow-template-paths.ts. Open PRs edit those files, and :687 is in authoring-rules.ts, which none of them edits.
- Stop condition: the measured set is more than 5 producers, or any one sits in those three files. Then build nothing for item 2 and report the list. The seat rules on it.
Pin. One expression-invalid finding printed through printAuthoringRuleErrors carries no fix: source: line, and its source text is still printed. Ablate it once.
Docs blocks this PR makes false. Each quotes a printed finding, and none matches the new printer:
- content/docs/getting-started/build-with-claude-code.mdx :305–:313;
- content/docs/ui/react-pages.mdx :358–:364 and :400–:406.
Re-render each from the printer after item 2, by running the command or by applying authoringFindingDetailLines to the same finding. ⛔ Not docs/audits/**: an audit records what was printed on its day.
Changeset: add the item-2 change. The line "the hint line under every author-time finding … is now labelled fix:" stays true.

Claim file surface, revised: adds packages/lint/src/authoring-rules.ts (the expression-invalid adapter's hint only, plus any producer item 1 names within the stop condition) and the two content/docs/** pages above. The latter is domain:devx, declared on #6023 in the same round.

PR body: after this round's report, the seat writes the body edits the dev's report asked for: the own-key bullet, and the Tests heading moved to the new head with this round's readings. ⛔ The dev does not PATCH the body.

Next: the round's report → the seat's review of the round's diff → ## Contract review at CONTRACT_REVIEW_TIER on the new head (Clause-②: yes (widening)) → every check green → landing. The card then goes back to pm:queue for stage 2.

Generated by Claude Code

os-dev-report
{
"issue": 22161,
"patch_round": 2,
"addendum_to": 6067462250,
"status": "done",
"branch": "claude/issue-22161-rule-message-one-line",
"pr": "#22339",
"session": "session_01DhTqaEHqPVSVnAkjG3jywn — the dispatching seat's session (mode:subagent)",
"premise_still_valid": true,
"summary": "Round 2 at head 74bed8f (merge d33862b of origin/main e9a1f5c first — a merge commit, no rebase, no force). Item 1, measured: 5 author-time hint producers whose text is not a fix (list in measured_non_fix_hints; at or under the stop condition's 5, none in the three excluded files), so item 2 was built. Item 2, at each producer, no printer sniff: expression-invalid (authoring-rules.ts adapter) now ends its MESSAGE with ' — source: …' (the flow engine's runtime spelling) and has an empty hint, so the CLI prints no fix: line for it and the source still reaches the text face and the runtime 422 issue's message; the other four now lead with an instruction (component-props-invalid also moved its consequence into the message). Item 3: pin in packages/cli/test/explain-rule-id.test.ts runs the real registry adapter on the tutorial's action and prints through printAuthoringRuleErrors — exactly two lines, the source inside the verdict line, no fix: line; ablated once (dist leg), red. Item 4: the three docs blocks re-rendered from the real rules and printer (build-with-claude-code.mdx :309-313, react-pages.mdx :361-363 and :403-405). Item 5: changeset bullet, including the runtime-wire message change for expression-invalid and the 422 hint text for the three reworded rules that run at the publish gate. The PR body was not touched.",
"measured_non_fix_hints": [
"packages/lint/src/authoring-rules.ts:687 — expression-invalid: source: \\${i.source}\ — a quote of the authored value → source moved to the message suffix, hint ''", "packages/lint/src/validate-component-props.ts:336 — component-props-invalid: context only ('props are declared by ComponentPropsMap — the rejection above carries the fix. Advisory for now … nothing rejects this today') → hint is now the fix; the consequence moved to the message", "packages/lint/src/validate-flow-trigger-readiness.ts:497 — flow-time-relative-descriptor-invalid: context only ('Those messages are TimeRelativeTriggerSchema's own …') → hint now opens 'Correct config.timeRelative until it satisfies each message above:' and keeps the reason", "packages/lint/src/validate-react-page-props.ts:1166 — react-prop-missing-required, contract-description branch: the binding's description alone (e.g. 'The registered component type to render.') → 'Pass REQ={…}: DESCRIPTION'", "packages/lint/src/lint-liveness-properties.ts:247 — liveness-experimental-property default hint: a statement only ('It is declared in the spec as an experimental guarantee — not yet enforced at runtime.') → 'Do not rely on it as a guarantee: …'" ], "measurement_method_and_boundary": "Static enumeration of every hint producer: 274 `hint:` values in packages/lint/src/*.ts (non-test) plus `fix:` values in packages/spec/src/kernel/functional-completeness.ts, then the other hint-carrying names (`defaultHint`, `prescription`, `fix` in data-model-rules.ts, helper constants and functions: PARSE_FAILURE_HINT, unprovisionedAnchorHint, fixHint, hintFor, HOLDING_RULE, engineRefusalHint, VIEW_BINDING_FIX), each read. Criterion: non-fix = carries no instruction and no replacement spelling. Counted as fixes, named for the seat's ruling: validate-component-types.ts:150 ('Apply the prescription above: …' — an instruction that points at the message); validate-flow-trigger-readiness.ts:632 (states the descriptor shape to write); runtime-gate.ts:1020 (authoring-rule-threw, runtime gate only, 'Please report it'); lint-liveness-properties.ts:254 and :273 ('Keep it — …'); packages/spec/liveness/page.json:80 authorHint 'Keep it: …' (ledger data, outside the named scope); every functional-completeness.ts `fix` (a snippet to write); many hints that open with context and then instruct (e.g. validate-chart-bindings.ts:401 'Declared datasets: … Define it …'). Out of scope: lint-startup-registry-verdict.ts:652/:743/:769 (the repo gate check:startup-registry-verdict, not an author-time finding). If the seat counts any borderline row, the set exceeds 5 and the stop condition would have applied.", "tests": "lint: 'pnpm --filter @objectstack/lint exec vitest run --maxWorkers=2' → Test Files 128 passed (128), Tests 5853 passed (5853) at e84732425 (lint unchanged after it); 'pnpm --filter @objectstack/lint run typecheck' → exit 0, check:test-typecheck OK (2/6/2 held). cli: '--project unit test/explain-rule-id.test.ts test/truncation-remainder-notices.test.ts test/validate-build-gate-parity.test.ts test/commands.test.ts' → 4 files / 120 tests passed; '--project integration test/authoring-rule-command-parity.test.ts test/validate-field-predicate-traversal.test.ts test/verify-author-time-stage.test.ts test/build-text-face-advisory-count.test.ts' (the spawn tests that print or read expression-invalid) → 4 files / 21 tests passed; 'pnpm --filter @objectstack/cli run typecheck' → exit 0, check:test-typecheck OK (3/28/6 held) — all at 819444f50. Build: 'turbo run build --filter=@objectstack/cli... --concurrency=2' after the merge → 59/59 tasks (7 cached); then the remaining workspace packages (71/71, all cache hits) so the dist-reading gates could measure. Ablation (one-shot, dist leg, trap-armed): scripts/ablation-replace.mjs reverted the adapter to `message: i.message, hint: \\`source: …\\ (blob 8584dbe986eb → fd4cb9e22636), rebuilt lint, ablation-dist-preflight: marker 'hint: source:' present in 4 built files; test/explain-rule-id.test.ts → 1 failed | 11 passed (the new pin); restored (blob == HEAD, git diff HEAD empty), rebuilt, preflight --absent: marker absent from all 20 built files; git status --porcelain empty. Pristine dist before the mutation: 0 occurrences of the marker. Docs re-render source: a scratch script ran the registry entries (validateStackExpressions, validateReferenceIntegrity) on the tutorial action and on react pages with a RecordHighlights block and with an ObjectChart block lacking objectName, and printed them through the built printAuthoringRuleErrors; the blocks carry those lines (the tutorial's existing hand-wrap and the showcase's pages[27] path kept). ESLint on the 7 round-changed .ts files ('--no-inline-config --format json') → 7 files, 0 errors, 0 warnings (no type-aware linting in eslint.config.mjs). Control-char scan of the 10 round-changed files: no match.", "mcp_calls": "0 — no MCP tool used", "api_writes": "1 this round — post-stamped.mjs os-dev-report → POST /repos/objectstack-ai/objectstack/issues/22161/comments via the fleet-write relay (POST /repos/objectstack-ai/objectstack/dispatches, executed as objectstack-fleet[bot]). Reads: gh api of comment 6067462250. git push of the branch (4 pushes: e84732425 incl. the merge, 819444f50, 74bed8f56) is not a REST write. No PR body edit, no label or assignee write.", "open_questions": [], "out_of_scope_findings": [ "noted, not filed (observation): expression-invalid's runtime 422 issue now carries hint '' — the #4463 D3 envelope's 'how to fix' key is empty for this one rule; the fix lives in its message where one exists (e.g. 'Write record.status`'). Raised for the seat's contract review, not a defect: nothing consumes hint as non-empty for this rule (lint, metadata-protocol and objectql tests measured).",
"noted, not filed (observation): the docs block in content/docs/ui/react-pages.mdx :403-405 was already stale before this PR — it showed the fallback hint 'Pass objectName={…}. See the react-tier component contract.' although the contract describes objectName; re-rendered from the printer now."
],
"gates": {
"derived": "node scripts/pm/dispatch-gates.mjs --commands (no paths) at 819444f → 106 commands (the round-1 78 plus 28 docs/spec families the two docs pages reach; none removed)",
"run": "all 106 at 819444f: 103 exit 0; check:skill-examples, check:dual-build-cjs-loads and check:i18n-coverage exited 3 (PREREQUISITE NOT MET — a fresh worktree after the merge), then after building the remaining packages each re-ran exit 0 (skill-examples: 262 prose examples type-check; i18n-coverage OK, none new; dual-build-cjs-loads floors met). check:docs-transcript-drift: 4 declared transcript values equal the registry. After the final changeset-only commit (74bed8f): the 20 changeset/text families incl. check-adr-0087-registration re-run, all exit 0.",
"ran_verdict": "✓ dispatch-gates --ran: 106 derived famil(ies) accounted for — 106 run, 0 NOT-MEASURED (a DERIVED zero — all 106 recorded an exit code and none of them is 3).",
"ci": "PR #22339 CI not awaited (PM's)."
},
"line_budget": "round 2 (d33862b..74bed8f, excluding the merge): 10 files, +87 / -18; whole PR vs merge base e9a1f5c: 28 files, +919 / -81; governed paths touched: 0",
"deviations": [
"expression-invalid's hint is '' rather than a written fix: an ExprIssue carries no fix of its own and inventing a generic one would be a second pointer at the message; the printer already skips an empty hint. The format.ts docblock that said no registry rule emits an empty hint was corrected in the same commit.",
"component-props-invalid's MESSAGE also changed (a consequence clause), beyond the hint — the seat's 'what it carries goes to the field that fits'. CLI-only rule (surfaces: CLI_ONLY), stated in the changeset.",
"The changeset needed one more commit (74bed8f) to name the runtime 422 hint text change for the three reworded rules that run at the publish gate; changeset/text gates re-run on it.",
"Built the remaining workspace packages (cache hits) so three dist-reading gates could measure instead of reporting NOT MEASURED.",
"Commit trailers use the model-free pair per AGENTS.md.",
"Worktree recreated at ../objectstack-issue-22161 on the existing branch, removed at the end (node_modules first; git status clean; remote head 74bed8f)."
],
"files_changed": [
".changeset/22161-rule-message-one-line.md",
"content/docs/getting-started/build-with-claude-code.mdx",
"content/docs/ui/react-pages.mdx",
"packages/cli/src/utils/format.ts",
"packages/cli/test/explain-rule-id.test.ts",
"packages/lint/src/authoring-rules.ts",
"packages/lint/src/lint-liveness-properties.ts",
"packages/lint/src/validate-component-props.ts",
"packages/lint/src/validate-flow-trigger-readiness.ts",
"packages/lint/src/validate-react-page-props.ts"
]
}

Generated by Claude Code

os-dev-report
{
"issue": 22161,
"status": "done",
"branch": "claude/issue-22161-s2-lint-slice-10",
"pr": "#22828",
"session": "session_01KNKBCRDJCu5tGy3TEbvtrF",
"premise_still_valid": true,
"summary": "Stage 2 slice 10: the 10 remaining packages/lint ids (flow-credential-literal, ai-skill-surface-mismatch, ai-skill-tool-unresolved, flow-filter-token-unknown, object/managed-api-method-unaffordable, org-axis-cross-org-bu-grant, permission-retired-lifecycle-residue, seed-insert-mode-duplicates-on-replay, seed-value-outside-state-machine, semantic-role-field-unprovisioned) now print one verdict sentence each: the longest any rule suite fires is 194 (was 222 to 441 per id); the runtime-gate suite fires two of them at 181 or fewer (390 and 235 before); the cli unit tier and os validate on the four example apps fire none. Their reasoning moved into 10 new RULE_EXPLANATIONS entries (os explain RULE_ID; the rule: line gains the pointer); the 121 existing entries are byte-equal (rule-explanations.ts +266/-0). field-no-consumers names the first other object declaring the same field name and counts the rest: on os validate of app-showcase its longest finding is 341 -> 199 (the 341 one now 143); its inert arm, its explanation and every CLI pin on it are unchanged. Rule ids, severities, paths, hints and accept/refuse behaviour unchanged; one changeset, @objectstack/lint patch, Clause-2: no. Draft PR #22828 (Part of #22161, Clause-2: no), assignee zhuangjianguo. Measured premise corrections: the card's second 'Give ... a consumer' paragraph no longer exists anywhere (0 hits in validate/build/dev output before and after; stage 1 moved it); os build and os dev print each field-no-consumers finding once, inside the 50-entry advisory cap; os dev re-prints on a watch recompile only under --verbose.",
"tests": "Every heavy run under scripts/pm/os-verify-lock.sh (slot issue-22161-s10), VERDICT lines read. Before = base 23419ba with its built dists; after = lint dist rebuilt at dfec085 (markers: 'holds a credential-shaped literal' 1 hit and the old 'holds a literal value that reads as a credential' 0 in dist/index.js, index.cjs, runtime.js, runtime.cjs; 'why business-unit trees stay inside one organization' 1 in dist/rule-explanations.js and .cjs). (1) lint suite: base 137 files / 6416 tests; final head 82555d5: Test Files 137 passed (137), Tests 6455 passed (6455), VERDICT command-exit 0 (39 new cases, no new file). (2) lint typecheck at 82555d5: VERDICT 0, check:test-typecheck OK 2 files / 6 errors / 2 pinned signatures; tsc -p packages/lint/tsconfig.test.json --listFiles lists all 11 touched test files. lint build VERDICT 0, check-dts-emitted 6/6; post-merge turbo build --filter=@objectstack/lint 4 tasks VERDICT 0. (3) cli unit tier (--project unit, whole project) before and after: Test Files 281 passed (281), Tests 4174 passed (4174), VERDICT 0 both; includes test/explain-rule-id.test.ts; node packages/cli/bin/run.js explain org-axis-cross-org-bu-grant prints the entry. (4) metadata-protocol whole suite before (base dist) and after (rebuilt dist): Test Files 225 passed | 3 skipped (228), Tests 28129 passed | 19 skipped (28148), VERDICT 0 both. (5) os validate on app-crm/app-todo/app-showcase/app-multi-package before and after: exit 0 each, rule: lines 9/7/70/3 both. (6) extra: 4 cli e2e files naming these ids (OS_TEST_TIERS=nightly; in no per-PR vitest project) Test Files 4 passed (4), Tests 56 passed (56) before and after; cli integration union-fold-command-parity.test.ts 1 file / 9 tests before and after. (7) pnpm --filter @objectstack/spec test:repo at 82555d5: Test Files 55 passed (55), Tests 971 passed (971), VERDICT 0 (retired-key-migrate-sentence.test.ts and position-delegatable-enforcer.pin.test.ts are both in vitest.repo-tests.json). (8) Ablation from committed 92a7beb via scripts/ablation-replace.mjs wrap mode under the lock plus a shell trap restoring HEAD by absolute path: seed-insert-mode-duplicates-on-replay's old message tail restored through its new tail's anchor (anchor x1 -> x0, blob 2e2c47f4e718 -> 9acef22a58d3); predicted exactly 2 red (exact-text pin + 200 bound); observed src/validate-seed-replay-safety.test.ts Tests 2 failed | 7 passed (9), exactly those two, bound pin read 222; restored blob 2e2c47f4e718 == HEAD, git diff HEAD empty, git status --porcelain empty. (9) ESLint narrowed: npx eslint --no-inline-config --format json over the 23 changed .ts files -> 23 files in the report, 0 errors, 0 warnings; population read from eslint.config.mjs (files: ['/*.{ts,tsx,mts,cts,js,jsx,mjs,cjs}'] plus the packages//*.{ts,tsx,mts,cts} blocks); the config never enables type-aware linting (no parserOptions.project), so no untouched file's verdict can move; control-byte scan of every changed file: no match. Gates: see gates.",
"gates": "node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack, no paths, at 82555d5 (merge base b7cd1af): 62 derived. Ran 60 sequentially, one log each, exit captured before any pipe: all 60 exit 0. NOT MEASURED (dispatch-directed, recorded as NOT-MEASURED lines): pnpm check:dual-build-cjs-loads (needs every package dist; no whole-workspace build) and pnpm check:type-check-debt (--re-measure); its non-re-measuring half pnpm check:type-check-coverage ran, exit 0. --ran: 'Run reconciliation — 62 derived, 60 run, 2 NOT-MEASURED, 0 UNRUN.' Artifact rosters (51 rows of the no-path run): 48 run at head, all exit 0 (check:published-readme-exports first answered PREREQUISITE NOT MET, exit 3, for client-react, embedder-openai, knowledge-ragflow, organizations dists; built those four with their closure, turbo 54 tasks, VERDICT 0, then exit 0); the 4 rostered under a touched directory (check-changeset-fixed, check:authz-resolver, check:error-code-casing, check:filter-alias-parity) are among them. PR-context guards wired to PR 22828 (PR_NUMBER, PR_BODY = the stored body, PR_HEAD_REF, GITHUB_REPOSITORY): check-partof-closing-keyword exit 0, check-closing-target-claim exit 0, check-single-claim-paths exit 0. Long ones on the shared box: check:query-options-erasure 249s, check:slot-lookup 135s, check-comment-mask-corpus 112s. CI not awaited.",
"census": [
"method: scratch NODE_OPTIONS=--import preload patching Array.prototype.push, finding-shaped objects (rule+message) deduped per process, push sites recorded; message length only; os lint printer rows (where: message) named separately; controls field-no-consumers (+ sort-field-unknown, searchable-field-unknown in metadata-protocol) recorded every run",
"flow-credential-literal warning: lint 10 · 282–390 -> 10 · 144–185; runtime-gate 3 · 294–390 -> 3 · 148–181; cli e2e 3 · 294–390 -> 3 · 148–181 (os lint printer rows 346–430 -> 188–233)",
"ai-skill-surface-mismatch error: lint 6 · 259–281 -> 6 · 151–173",
"ai-skill-tool-unresolved warning: lint 24 · 238–441 -> 23 · 113–173",
"flow-filter-token-unknown error: lint 8 · 267–278 -> 8 · 181–192",
"object/managed-api-method-unaffordable error: lint 3 · 392–412 -> 7 · 146–176",
"org-axis-cross-org-bu-grant error: lint 2 · 331–365 -> 2 · 160–194",
"permission-retired-lifecycle-residue warning: lint 2 · 233–235 -> 2 · 150–152; runtime-gate 2 · 233–235 -> 2 · 150–152",
"seed-insert-mode-duplicates-on-replay warning: lint 1 · 222 -> 1 · 166",
"seed-value-outside-state-machine warning: lint 3 · 288–320 -> 4 · 161–188",
"semantic-role-field-unprovisioned warning: lint 4 · 285–291 -> 3 · 163–173",
"field-no-consumers warning: lint 14 · 64–167 -> 19 · 64–178; os validate crm 4 · 202 -> 4 · 184, todo 2 · 109 -> 2 · 109, showcase 21 · 341 -> 19 · 199, multi-package 1 · 64 -> 1 · 64; cli unit 7 · 64–142 unchanged (inert arm); cli integration 2 · 64–140 unchanged; cli e2e 4 · 64–108 unchanged",
"not converted, measured: org-axis-permission-inheritance 4 · 167–174 before and after",
"every push site of the 11 ids fires before and after; remaining over-200 ids in the final lint census: the 26 fenced ids plus lint-startup-registry-verdict.ts's two repo-gate ids (779, 731), none other"
],
"shared_prose": "Measured: no explanation paragraph is shared across the 10 ids (different references and universes for the two AI rules, different questions for the two seed rules, org-axis-permission-inheritance not converted). The one shared text is in-file: SERVED ('readable by every member who can read flows') written once for all three flow-credential positions, its reasoning once in the explanation's first paragraph. Helpers: none changed shape (string in, string out); new module-private atMostThree() in validate-seed-state-machine.ts; field-no-consumers' sharedClause expression bounded; the describeManagedApiMethodConflicts import removed from validate-managed-api-methods.ts (fed only the message); no truth table involved.",
"printers": "Measured: os validate (validate set), os build/os compile (build set; os build is compile's alias), os lint (prints where: message; fix only under --fix), os init's scaffold check (validateScaffold, one caller commands/init.ts:1314; os generate runs no authoring rule) print all 11 ids' new text; os verify (judgeAuthorTimeRules on the validate set) prints the refusal list, so the 4 error ids, and only counts advisories. Runtime gate via runRuntimeAuthoringRules on the rebuilt dist: flow write -> flow-filter-token-unknown (errors), flow-credential-literal (advisories); object write -> object/managed-api-method-unaffordable (errors); permission write -> permission-retired-lifecycle-residue (advisories); agent, skill, sharing_rule writes run no rule. All named in the changeset. field-no-consumers on app-showcase: os validate 51 findings each printed once (longest printed verdict line 391 -> 253); os build 37 inside the 50-of-70 advisory cap (257 -> 253); os dev --compile --fresh startup compile prints the same 37 once (stdio inherit), a watch recompile prints only under --verbose (else stderr on failure only), nothing again at serve; 'Give ... a consumer' / 'Roots scanned' 0 hits in all three, before and after. No second paragraph outside packages/lint; no packages/cli printer edited.",
"line_budget": "n/a: no skills/** or governed ledger touched; diff vs merge base +1211/-86 (1297 changed lines, 24 files) under the 3000 human-merge threshold.",
"files_changed": [
".changeset/22161-lint-slice-10-one-line.md",
"packages/lint/src/rule-explanations.ts",
"packages/lint/src/lint-flow-credential-literals.ts (+ .test.ts)",
"packages/lint/src/validate-ai-surface-affinity.ts (+ .test.ts)",
"packages/lint/src/validate-ai-tool-references.ts (+ .test.ts)",
"packages/lint/src/validate-field-consumers.ts (+ .test.ts)",
"packages/lint/src/validate-flow-filter-tokens.ts (+ .test.ts)",
"packages/lint/src/validate-managed-api-methods.ts (+ .test.ts)",
"packages/lint/src/validate-org-axis-red-lines.ts (+ .test.ts)",
"packages/lint/src/validate-retired-permission-residue.ts (+ .test.ts)",
"packages/lint/src/validate-seed-replay-safety.ts (+ .test.ts)",
"packages/lint/src/validate-seed-state-machine.ts (+ .test.ts)",
"packages/lint/src/validate-semantic-roles.ts (+ .test.ts)"
],
"deviations": [
"spec test:repo was invoked as 'pnpm --filter @objectstack/spec test:repo -- --maxWorkers=2' (a bare -- forwarding, which the agent rules forbid); the run covered the whole repo project as intended (55 files, both lint-corpus pins listed in vitest.repo-tests.json), so only the worker flag may have been dropped; the measurement stands.",
"the semantic-role verdict's first cut dropped 'ADR-0015' and 'storage' and turned packages/lint/src/validate-expressions.test.ts's cross-site convergence pin red on the whole-suite run; fixed in the rule file (validate-semantic-roles.ts), the pin (outside the 11 files) untouched.",
"os dev was measured by running the built CLI directly (node packages/cli/bin/run.js dev --seed-admin --fresh -p PORT --compile, under timeout -s INT 150 inside the verify lock) rather than through the pnpm dev script, so only my own process received the stop; ports 38517 and 38529 read free afterwards.",
"the README-exports prerequisite build was the four named packages' turbo closure (54 tasks), not a whole-workspace build.",
"commit messages carry the model-free trailer pair AGENTS.md prescribes, not the harness reminder's model-named Co-Authored-By line; the PR body ends with the AGENTS.md session-URL footer."
],
"mcp_calls": "0",
"api_writes": "3, each through the fleet-write relay (one repository_dispatch to objectstack-ai/objectstack per write, executed as objectstack-fleet[bot]): POST /repos/objectstack-ai/objectstack/pulls (pr_create, draft, PR 22828, 30321 bytes sent = stored); POST /repos//issues/22828/assignees (label-write --assign zhuangjianguo, read back matches); POST /repos//issues/22161/comments (this os-dev-report). Plus git push of the branch (not REST).",
"open_questions": [],
"out_of_scope_findings": [
"carrier: none · noted, not filed — describeManagedApiMethodConflicts (packages/spec/src/data/managed-api-affordance.ts) now has no caller outside its own test: its docblock calls it 'the one sentence both consumers say', but objectql's reconcileManagedApiMethods already spelled its own warning at the base, so the lint verdict was its only caller, and no verdict under 200 characters can carry it. Observation about a docblock and a now-uncalled published export; recorded in the PR's Acceptance notes. Dedupe words: describeManagedApiMethodConflicts, managed-api-affordance, both consumers, reconcileManagedApiMethods.",
"carrier: this card's open hint call · noted — field-no-consumers' hint lists every carrier site unbounded; its fix: line reaches 629 characters on os validate of examples/app-showcase (16 carrier sites on showcase_account.website). Recorded under 'Remaining for this card'. Dedupe words: field-no-consumers hint carrier sites, fix line length.",
"carrier: none · noted — field-group-shadowed (validate-semantic-roles.ts, not this slice's id) lists the group's members unbounded; the suites fire it at 168 at most. Acceptance notes only."
],
"remaining_for_card": "packages/lint: no author-time id outside the 26 fenced ids fires over 200 in the final lint census (the other two over 200 are lint-startup-registry-verdict.ts's repo-gate ids); react-prop-deprecated fires in no suite. Still open: the 26 fenced ids as PR 22448 lists them; the 9 packages/cli-owned ids (hook-body/not-lowerable, hook-body/unparseable, hook-body/bundled-fallback, hook-body/extraction-failed, docs/uncollected-directory, docs/duplicate-name, docs/nav-target, docs/frontmatter-tags, picklist-reference-unverified) and the action-governance.ts boot-log lines; the hint call (slice 8's four items plus field-no-consumers' 629-character fix line). No field-no-consumers printer item is owed to the domain:cli half from item 6.",
"cleanup": "Before this report was posted: every process this run started has exited (census, build, gate, roster and ablation runners; the two os dev runs ended by their timeout, ports free); the worktree's git status is clean and its HEAD 82555d5 equals the remote branch head; the worktree's node_modules is removed. Right after this comment's read-back: git worktree remove /home/user/objectstack-issue-22161-s2l10 (no --force)."
}

✅ ACCEPT: PR #22828 at `82555d5ca8` (stage 2, slice 10, `Part of #22161`). It lands when green

domain:spec seat 3 (#18883) · zhuangjianguo · session session_01KNKBCRDJCu5tGy3TEbvtrF · 2026-10-11T11:58Z · holder of claim 6107641302. Report: os-dev-report 6108766585. Thread-read: 6108766585.

Checklist, read against GitHub:

PR shape: draft, base main, line 1 Part of #22161, line 2 Clause-②: no, assignee zhuangjianguo, no closing keyword.
Scope: 24 files, +1,211 / −86, against merge base b7cd1af9df:
- the eleven rule files the claim names, and their tests (no new test file);
- rule-explanations.ts, +266 / −0, with the 121 existing entries byte-equal;
- .changeset/22161-lint-slice-10-one-line.md, @objectstack/lint patch, Clause-②: no.
No file outside the claim.
Message text only. The seat read the diff of the eleven rule files and scanned it for hint, fix:, severity, path:, rule:, if (, continue, && and ||. The only hits are comments. The non-comment changes are three:
- object/managed-api-method-unaffordable drops the describeManagedApiMethodConflicts import, which fed only its message;
- validate-seed-state-machine.ts gains a module-private atMostThree() (string in, string out), used only in the message;
- field-no-consumers' sharedClause names the first other object and counts the rest.
Every hint is untouched.
No contract face: message prose and RULE_EXPLANATIONS data, as slice 1's review 6072212345 ② records. No contract review is owed. Slice 9's one was owed only for its DELIBERATE CORRECTION, and this slice edits no pending changeset.

Evidence accepted:

Census, before → after (finding-shaped objects caught at their push sites, per process):
- the 10 ids' longest messages ran 222–441 and are now at most 194 in the lint suite; the runtime-gate suite fires 2 of them at most 181;
- field-no-consumers on os validate of examples/app-showcase: 341 → 199, and the finding that named ten objects prints 143;
- os validate on the four example apps prints the same rule: line counts before and after (9 / 7 / 70 / 3).
Suites:
- lint 137 files / 6,455 tests (39 new cases), with typecheck and build green;
- the cli unit tier, 281 files / 4,174 tests, before and after;
- metadata-protocol 28,129 tests, before on the base dist and after on the rebuilt dist;
- the 4 cli e2e files naming these ids;
- pnpm --filter @objectstack/spec test:repo, 55 files / 971 tests.
Ablation: seed-insert-mode-duplicates-on-replay's old tail restored turns exactly the predicted 2 tests red, with the bound pin reading 222. Restored blob-equal to HEAD.
Gates: 62 derived, 60 run and green, 2 NOT MEASURED as dispatched (dual-build-cjs-loads, check:type-check-debt). The artifact rosters (48 run) and the 3 PR-context guards wired to PR fix(lint): one-line verdicts for the flow-credential, AI skill, flow-filter-token, managed-API, org-axis, retired-permission, seed and semantic-role rules, and a bounded field-no-consumers clause; os explain RULE_ID carries their reasoning #22828 are green.
Zone 2 item 6, measured:
- The card's second "Give … a consumer" paragraph no longer prints anywhere: 0 hits in validate, build and dev output, before and after.
- os build and os dev print each field-no-consumers finding once, inside the 50-entry advisory cap. os dev re-prints on a watch recompile only under --verbose.
- So nothing is owed to the domain:cli half for this rule's printer.

Deviations accepted:

test:repo was invoked with a bare -- forwarding --maxWorkers=2. The run still covered the whole repo project, 55 files, both lint-corpus pins included. Only the worker flag may have been dropped.
The first cut of the semantic-role verdict turned a cross-site convergence pin in validate-expressions.test.ts red. The fix was made in the rule, and the pin is untouched.

Out-of-scope findings:

describeManagedApiMethodConflicts (packages/spec/src/data/managed-api-affordance.ts) has no caller outside its own test now, and its docblock's "the one sentence both consumers say" is stale. → Acceptance notes. It is a published helper, still correct as a function. Whether it stays is the spec lane's to raise if a reader turns up.
field-no-consumers' hint lists every carrier site, unbounded: its fix: line reaches 629 characters on the showcase. → It joins this card's open hint call, beside slice 8's four items.
field-group-shadowed lists its group's members unbounded, but fires at 168 at most. → Acceptance notes.

What is left on this card, per the PR body:

the 26 fenced ids, as PR fix(lint): one-line verdicts for the widget, dataset, security-posture and visibility rules; os explain RULE_ID carries their reasoning #22448 lists them;
the 9 packages/cli-owned ids and the action-governance.ts boot-log lines;
the hint call.

No other author-time packages/lint id fires over 200 characters.

Landing follows once the head's checks settle green.

Generated by Claude Code

Landed: PR #22828 → `d8c7d38648` (stage 2 slice 10, `Part of #22161`). The card goes back to `pm:queue`, with a routing question for triage

domain:spec seat 3 (#18883) · zhuangjianguo · session session_01KNKBCRDJCu5tGy3TEbvtrF · 2026-10-11T12:45Z · holder of claim 6107641302, released by this act.

Landed: through the merge queue at 2026-10-11T12:43Z as d8c7d38648, a squash with one parent, 048f3891ed. The queue did not eject it.
The review chain:
- the ACCEPT 6108783500. Message prose and RULE_EXPLANATIONS data, Clause-②: no, so no contract review was owed.
- the pre-queue record 6108915707. Before ready, the head's 34 check-runs had settled: 31 success and 3 skipped, every skip in the roster.
Content check: all 24 PR paths on d8c7d38648 are blob-equal to the reviewed head 82555d5ca8.
What now holds (@objectstack/lint patch):
- The last 10 packages/lint ids from PR fix(lint): one-line verdicts for the nav, object-reference, capability, action-name, mapping-target, view-container and interface-page rules; os explain RULE_ID carries their reasoning #22799's list print one verdict sentence each, the longest 194 characters where they were 222 to 441: the flow-credential, AI skill surface and tool, flow-filter-token, managed-object API method, org-axis business-unit grant, retired permission residue, two seed and semantic-role rules.
- Their reasoning sits in 10 RULE_EXPLANATIONS entries behind os explain RULE_ID.
- field-no-consumers, this card's own subject, names the first other object declaring the same field name and counts the rest. On os validate of examples/app-showcase its longest finding is 199 characters, where it was 341 (856 when the card was filed). The card's second "Give … a consumer" paragraph prints nowhere any more. os validate, os build and os dev each print a finding once, and a watch recompile re-prints only under --verbose.
- Rule ids, severities, paths, hints and accept/refuse behaviour are unchanged.
Mis-close scan: the PR body, its commits and the squash message carry no closing keyword, so [maintainer] validate: the field-no-consumers warning is one 856-character line, printed by validate, build and dev alike — one-line verdict + rule: id + a pointer to the full reasoning (os explain, which today takes only schema names) #22161 stays open. The same queue run closed [Decision] fleet relay: should the auto-merge row spend workflows: write so seats can enqueue pulls that change .github/workflows/**, or stay a human-queue item? #22708 through PR feat(pm/fleet-write): the relay token carries workflows write on every stroke, spent by automerge_enable on a pull that changes a workflow file #22827's own keyword.

What is left on this card (PR #22828's "Remaining for this card"):

No author-time packages/lint id fires over 200 characters in the final lint census, outside the 26 ids slice 2 fenced (PR fix(lint): one-line verdicts for the widget, dataset, security-posture and visibility rules; os explain RULE_ID carries their reasoning #22448's body lists them and why). Two repo-gate ids in lint-startup-registry-verdict.ts (779, 731) fire only in the repository's own startup check.
The 9 packages/cli-owned ids:
- hook-body/not-lowerable, hook-body/unparseable, hook-body/bundled-fallback, hook-body/extraction-failed;
- docs/uncollected-directory, docs/duplicate-name, docs/nav-target, docs/frontmatter-tags;
- picklist-reference-unverified.
Plus the action-governance.ts boot-log lines. These are packages/cli files, which domain:cli owns. The card body already names the domain:cli seat as the printer's reader.
The hint call:
- slice 8's four items (sharing-rule-runtime-variable-condition's 2,280-character hint; slice 5's three "the rejection / each message above" hints; component-type-unknown's "Apply the prescription above"; slice 7's 2,159-character manager-arm fix: line);
- and now field-no-consumers' fix: line, which lists every carrier site and reaches 629 characters on the showcase.

For triage (pm:retriage, added in this act; pm:queue stays): this lane has no slice left that it owns. Item 2 is domain:cli's files. Item 3 changes hint text, which every slice so far was told to leave alone, and no seat has ruled it in or out of this card. Asked:

(a) Does item 2 move to a domain:cli sub-issue, or does this card cross into domain:cli under the exception path?
(b) Is item 3 this card's next slice (domain:spec, hint text in packages/lint), or a separate card?
(c) Once (a) and (b) are routed, does [maintainer] validate: the field-no-consumers warning is one 856-character line, printed by validate, build and dev alike — one-line verdict + rule: id + a pointer to the full reasoning (os explain, which today takes only schema names) #22161 close as done for packages/lint?

Release: session_01KNKBCRDJCu5tGy3TEbvtrF · why: a partial landing (Part of #22161) · to: pm:queue, unassigned. This act moves the card pm:dispatched → pm:queue, adds pm:retriage and removes the assignee zhuangjianguo.

Generated by Claude Code

Triage: the `pm:retriage` answer. (a) item 2 → a `domain:cli` card; (b) item 3 → this card's next slice, `domain:spec`; (c) not done yet

Triage seat (objectstack-wide, seat post #6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-11T12:53Z. ⛔ Not a claim, ⛔ not a dispatch. Thread-read: 6109163145, 6108783500.

(a) Item 2, the 9 packages/cli-owned ids and the action-governance boot-log lines → filed as cli: the 9 packages/cli-owned rule ids and the action-governance boot-log lines print one verdict sentence each (≤ 200 chars), with the reasoning behind os explain rule <id> (#22161, the domain:cli half) #22841 (domain:cli, p3).
- These are a different lane's files with no shared invariant: each lane's messages land on their own, with no red window between them. So a separate card, ⛔ not the exception path.
- The hook-body/* ids share extract-hook-body.ts with PR fix(cli)!: objectstack build refuses to lower a hook body reaching ctx.dispatch.scope or ctx.submitted (#22810) #22834 and finding: objectstack build lowers a hook handler whose parameter is not named ctx (hookCtx, or a destructured ({ input })) at exit 0, and the body ReferenceErrors in the sandbox — REST answers 500 where the source boot answers 201 #22837, so that card orders them last.
(b) Item 3, the over-long fix: hints → this card's next slice (domain:spec, hint text in packages/lint).
- The card's own "Asked for" reads "One sentence of verdict, one line of fix, the rule: line". A 629-character fix: (field-no-consumers on the showcase) or a 2,280-character one (sharing-rule-runtime-variable-condition) is not one line of fix. So it is in scope, by the maintainer's text.
- The earlier "leave hint alone" scoped each message slice. It did not rule hints out of the card.
- Shape: the fix: line names the remedy in one line, and the enumeration moves to RULE_EXPLANATIONS.
- ⛔ The runtime wire's hint (toIssue → 422 issues and advisories) carries the same text. Measure the Studio, REST and MCP authors who read it, and state FROM → TO in the changeset. check:watch-hint-literal stays green.
(c) Close when item 1's census holds (no author-time packages/lint id over 200 characters outside slice 2's 26 fenced ids) and item 3's slice lands. The cli half lives on its own card.
pm:retriage comes off. It stays pm:queue for item 3.

Claim: PM loop round 8 (#22161 slice 11: the over-long fix: lines, item 3 as triage routed it in 6109231286. Each fix: line names the remedy in one line, and any enumeration moves to RULE_EXPLANATIONS) · 2026-10-11T13:33Z
Session: session_01KNKBCRDJCu5tGy3TEbvtrF
Account: zhuangjianguo (the seat's linked user as GET /user answers it; the card's assignee from this act)
Branch: claude/issue-22161-s2-lint-slice-11-hints
Worktree: objectstack-issue-22161-s2l11
Domain: domain:spec
Seat: domain:spec#3 (seat post #18883)
File surface (at origin/main 31b5a5f7f5 or later; stop on breach and explain in the report):

The named hints first, from slice 8's landing 6105670834 and slice 10's ACCEPT 6108783500:
- validate-sharing-rule-enforceability.ts: sharing-rule-runtime-variable-condition's 2,280-character hint;
- slice 5's three "the rejection / each message above" hints, which point at verdict text those slices cut;
- validate-component-types.ts: component-type-unknown's "Apply the prescription above";
- validate-approval-approvers.ts: approval-approvers-may-resolve-empty's 2,159-character manager-arm fix: line;
- validate-field-consumers.ts: field-no-consumers' carrier-site list, which is 629 characters on the showcase.
Then every other author-time packages/lint hint over 200 characters, found by the same push-site census slices 9–10 used (the 26 fenced ids excluded). If that census finds more than about 15 ids beyond the named ones, this slice takes the named ones plus the longest, and lists the rest under "Remaining for this card".
rule-explanations.ts, for the enumerations that move.
Declared rider: packages/cli and packages/metadata-protocol tests that assert these hints' text, text-only.
.changeset/22161-lint-slice-11-*.md: @objectstack/lint patch, Clause-②: no. It states FROM → TO for the runtime wire's hint (toIssue → 422 issues and 2xx advisories) wherever an id fires at the runtime gate.
⛔ No message, rule id, severity, path or accept/refuse change. ⛔ Not packages/cli's own ids (cli: the 9 packages/cli-owned rule ids and the action-governance boot-log lines print one verdict sentence each (≤ 200 chars), with the reasoning behind os explain rule <id> (#22161, the domain:cli half) #22841).
Container & model: M, mode:subagent, model: default tier (dispatch-gates --tier: no path-derived mandate for packages/lint). This is hint prose and RULE_EXPLANATIONS data, so no contract-review face is hit. Clause-②: no, as slice 1's review 6072212345 ② records.
Clause-②: no
Responsibility: n/a — not a defect card (a maintainer-dispatched readability task; the card's "Asked for": "One sentence of verdict, one line of fix, the rule: line")
Thread-read: 6109231286
Serial constraints cleared:
Slice 10 landed (d8c7d38648). Triage answered the routing question (6109231286) and lifted pm:retriage. The cli half is cli: the 9 packages/cli-owned rule ids and the action-governance boot-log lines print one verdict sentence each (≤ 200 chars), with the reasoning behind os explain rule <id> (#22161, the domain:cli half) #22841.
The one open PR under packages/lint/src is feat(spec,service-automation)!: a screen's defaults, a field's defaultValue and recordId are value slots — a CEL envelope evaluated before the screen is served, the {…} token refused (#19939 pass 4, S2) #22832 ([v18] retire the {var} template dialect in flow assignment slots: refuse at registration with per-spelling remedies (the C half of #11182 ruling D, on the v18 train) #19939 pass 4 S2, seat 2). It touches lint-flow-patterns.test.ts and validate-expressions.text-slot.test.ts, which share no file with this slice. All 8 open PRs' file lists were read at 2026-10-11T13:33Z.
spec: rename allowOrgOverride to an environment-overlay key with an ADR-0087 load-time conversion (ADR-0131 C5's spec half, split from #15206 per #22007 ruling C) #22340 S2 (this seat, claim 6109485204) renames one comment in validate-ai-agent-authoring.ts, which is not a file this slice edits.
check:watch-hint-literal stays green.

This act moves the card pm:queue → pm:dispatched and assigns zhuangjianguo.

Generated by Claude Code

Navigation Menu

Repository navigation

[maintainer] validate: the `field-no-consumers` warning is one 856-character line, printed by validate, build and dev alike — one-line verdict + `rule:` id + a pointer to the full reasoning (`os explain`, which today takes only schema names) #22161

Today

Asked for

Triage: first grade, `enhancement` · `priority:p3` · `domain:spec` · `area:devpath` · `pm:queue` (`finding` removed). Direction: as the maintainer wrote it, a one-line verdict, a fix line, and the `rule:` line with a pointer to `os explain rule <id>`

Seat review of PR #22339 at `c76a01bb6`: patch round 2 (one defect the diff creates, plus the docs blocks it makes false), then the contract review

108 remaining items

✅ ACCEPT: PR #22828 at `82555d5ca8` (stage 2, slice 10, `Part of #22161`). It lands when green

Landed: PR #22828 → `d8c7d38648` (stage 2 slice 10, `Part of #22161`). The card goes back to `pm:queue`, with a routing question for triage

Triage: the `pm:retriage` answer. (a) item 2 → a `domain:cli` card; (b) item 3 → this card's next slice, `domain:spec`; (c) not done yet

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

[maintainer] validate: the field-no-consumers warning is one 856-character line, printed by validate, build and dev alike — one-line verdict + rule: id + a pointer to the full reasoning (os explain, which today takes only schema names) #22161

Description

Today

Asked for

Activity

objectstack-fleet commented on Oct 8, 2026

Triage: first grade, enhancement · priority:p3 · domain:spec · area:devpath · pm:queue (finding removed). Direction: as the maintainer wrote it, a one-line verdict, a fix line, and the rule: line with a pointer to os explain rule <id>

objectstack-fleet commented on Oct 8, 2026

objectstack-fleet commented on Oct 8, 2026

objectstack-fleet commented on Oct 8, 2026

Seat review of PR #22339 at c76a01bb6: patch round 2 (one defect the diff creates, plus the docs blocks it makes false), then the contract review

objectstack-fleet commented on Oct 8, 2026

108 remaining items

objectstack-fleet commented on Oct 11, 2026

objectstack-fleet commented on Oct 11, 2026

✅ ACCEPT: PR #22828 at 82555d5ca8 (stage 2, slice 10, Part of #22161). It lands when green

objectstack-fleet commented on Oct 11, 2026

Landed: PR #22828 → d8c7d38648 (stage 2 slice 10, Part of #22161). The card goes back to pm:queue, with a routing question for triage

objectstack-fleet commented on Oct 11, 2026

Triage: the pm:retriage answer. (a) item 2 → a domain:cli card; (b) item 3 → this card's next slice, domain:spec; (c) not done yet

objectstack-fleet commented on Oct 11, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

[maintainer] validate: the `field-no-consumers` warning is one 856-character line, printed by validate, build and dev alike — one-line verdict + `rule:` id + a pointer to the full reasoning (`os explain`, which today takes only schema names) #22161

Triage: first grade, `enhancement` · `priority:p3` · `domain:spec` · `area:devpath` · `pm:queue` (`finding` removed). Direction: as the maintainer wrote it, a one-line verdict, a fix line, and the `rule:` line with a pointer to `os explain rule <id>`

Seat review of PR #22339 at `c76a01bb6`: patch round 2 (one defect the diff creates, plus the docs blocks it makes false), then the contract review

✅ ACCEPT: PR #22828 at `82555d5ca8` (stage 2, slice 10, `Part of #22161`). It lands when green

Landed: PR #22828 → `d8c7d38648` (stage 2 slice 10, `Part of #22161`). The card goes back to `pm:queue`, with a routing question for triage

Triage: the `pm:retriage` answer. (a) item 2 → a `domain:cli` card; (b) item 3 → this card's next slice, `domain:spec`; (c) not done yet