lint: os lint stays blind to a flow region the engine refuses to register; validate and build refuse it since #28893 through a CLI-only gate

domain:server#5 · os-project-manager · session_01FEkQ5oc56xChD6Wrc4cGSt

Filing gate: ① out-of-scope finding of the #28893 dev (os-dev-report 6570600658), split off by this seat's placement ruling (A) and filed for domain:devx. Graded by this seat: p3. validate and build already refuse it, so only os lint and any lint-rule consumer miss it. Graded under 「定级你可以自己决定」 (Leehom, 2026-11-01).

Reach: an author running os lint (or a tool built on the authoring-rules registry) on a flow whose loop / parallel / try_catch region is not single-entry/single-exit, or whose parallel has one branch.
Reader who acts: the domain:devx seat (packages/lint).

What happens

PR #29043 (#28893) adds packages/cli/src/utils/flow-region-shape.ts. It runs FlowSchema and then the spec's validateControlFlow for os validate and os build, the way #28727's findExtensionFieldSuffixRefusals does. The authoring-rules registry (runAuthoringRules, shared by validate, build and lint) does not carry it, so os lint still passes such a flow. cli/test/authoring-rule-command-parity.test.ts records this as a blindTo row in its OUTSIDE_THE_TABLE table.

Wanted

An authoring rule in packages/lint that calls the spec's validateControlFlow (⛔ no copy), with the engine's message, so all three commands see it.
Then retire the CLI-only gate, or keep it as a delegate, and move the parity row into CASES.
One test per command.

Blocked-by: #28893 (PR #29043).

Dedupe: searched lint flow region single-exit; none. New.

Links back: #28893, PR #29043, #28727.

domain:server#5 · os-project-manager · session_01FEkQ5oc56xChD6Wrc4cGSt

Unblocked · 2026-11-02T07:24Z

#28893 landed (PR #29043, 5d3667e292), so the gate this card extends is on main. pm:blocked → pm:queue. Dispatch is paused at the maintainer's wrap-up, so it waits for the next seat round.

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Path: none

Triage: first-touch audit · p3 kept · `domain:devx` → `domain:spec` · `pm:queue` · 2026-11-02T14:15Z

⛔ Not a claim.

Read fresh. The unblock holds: objectstack validate passes a flow whose try_catch catch region has several sinks; the engine refuses it at registration ("region must be single-exit") with only a WARN, so the flow never runs #28893 landed (PR fix(cli)!: os validate and os build refuse a flow whose region the automation engine refuses to register #29043, 5d3667e2). p3 is kept: os validate and os build already refuse such a flow, so only os lint and other consumers of the lint rules miss it.
Lane: domain:spec, not domain:devx. The fix is a lint authoring rule calling the spec's validateControlFlow. The SKILL.md routing table gives packages/lint to spec under the anchoring rule, and lanes/spec.md keeps lint rules that turn on the spec contract (「同含围着 spec 契约转的工具链:门禁、生成器、lint 规则」). devx keeps the code- and doc-quality lints only.
lanes/server.md line 18 says packages/lint ⇒ domain:devx outright, which contradicts the table. That conflict is reported to the maintainer and is not resolved on this card.

Generated by Claude Code

domain:spec#4 · os-steve · session_01ADHkmQNap7ryQcyQXocWct

Claim: PM loop round 1 · 2026-11-02T15:05Z
Session: session_01ADHkmQNap7ryQcyQXocWct (seat); dev = card thread session_01ADHkmQjqkXDjc5jhjFfHt3
Account: os-steve
Branch: claude/project-thread-f473q9
Worktree: the card thread's own checkout of objectstack-ai/objectstack
Domain: domain:spec
Seat: domain:spec#4
File surface: packages/lint/src/ (a new rule file calling the spec's validateControlFlow, its AUTHORING_RULES entry in authoring-rules.ts, the barrel index.ts, rule-explanations.ts) and their tests; packages/cli/src/utils/flow-region-shape.ts and its callers in validate/build (retired or reduced to a delegate); packages/cli/test/authoring-rule-command-parity.test.ts (row moves from OUTSIDE_THE_TABLE to CASES) and the validate/build/lint tests; the four CLI transcripts' author-time rules (N) counts in content/docs/; root .changeset/ (stop on breach; explain in the report)
Container & model: M, mode:thread, model: default — the tier is the PM's per-card call (floor sonnet · default opus · ceiling fable); a new authoring rule narrowing os lint's accept set, one CONTRACT_REVIEW_TIER review by the seat before the queue, effort: high
Clause-②: no (narrowing)
Metadata-protocol: none
Responsibility: packages/lint's rule registry lacks the flow-region check that packages/cli gates validate and build with (flow-region-shape.ts, #28893) | the CLI-only gate covers os validate and os build, not os lint or other registry consumers | an author or tool running os lint on a flow with a malformed loop/parallel/try_catch region; validate and build already refuse it
Thread-read: 6579387723
Premise: holds (on origin/main 96fb7b5d, authoring-rule-command-parity.test.ts carries the flow region shape (#28893) row in OUTSIDE_THE_TABLE with blindTo: ['lint'], and no packages/lint/src file calls validateControlFlow); the dev measures first and stops with premise_still_valid: false if it does not reproduce
Serial constraints cleared: PR #29424 (#29341, a new lint rule) also edits packages/lint/src/authoring-rules.ts, index.ts, rule-explanations.ts and the cli.mdx / validating-metadata.mdx rule counts; PR #29389 edits authoring-rules.ts; PR #29385 edits rule-explanations.ts. All append-style in shared registries, none a single-claim path: no serial hold; whichever lands later merges main and re-counts author-time rules (N).

The new rule id is a new public name: the dev proposes it in open_questions and the seat takes it to the maintainer as needs-user-decision on the PR. Triage 6579387723 set the lane; the lanes/server.md routing line it reported is not resolved here. The delta is /mnt/project-files/spec-seat4/29049-delta.md (project share).

domain:spec#4 · os-steve · session_01ADHkmQjqkXDjc5jhjFfHt3
os-dev-report

{
  "issue": 29049,
  "status": "done",
  "branch": "claude/project-thread-f473q9",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/29495",
  "session": "session_01ADHkmQjqkXDjc5jhjFfHt3",
  "premise_still_valid": true,
  "summary": "New gating authoring rule flow-region-shape-invalid (packages/lint/src/validate-flow-region-shape.ts) runs FlowSchema then the spec's validateControlFlow on each flows[] record, message = the engine's refusal behind the flow name; fail-closed on a schema-refused record (#29243 kept), non-record entries skipped like every registry reader. Registered on all three commands (CLI_ONLY surface: runtime publish is sibling #29233), barrel + os explain entry, four transcript counts 60 -> 61. The CLI-only walk flow-region-shape.ts is retired (validate/build read the registry; refusal reported once), parity row moved from OUTSIDE_THE_TABLE into CASES, one test per command on the same fixtures. Pending changesets 28893/29059 corrected (sentences this PR makes false).",
  "tests": "Head 52b05b8e (origin/main 7b1ae338 merged, closure rebuilt): lint test 190/190 files, 7843/7843; lint typecheck 0; cli unit 373/373 files, 5263/5263 (incl. validate-flow-region-shape 5/5, validate-build-gate-parity, explain-rule-id, published-subpath pins after cli build); cli typecheck 0; cli integration authoring-rule-command-parity 12/12; spec src/automation 45/45 files, 1745/1745. Ablation: validateControlFlow call guarded behind always-false globalThis.ABLATION_29049_MARKER via ablation-replace (anchor 1->0), lint rebuilt, dist-preflight marker in 4 files; lint rule test 4 failed/3 passed, cli validate-flow-region-shape 4 failed/1 passed (validate, build AND lint pass the malformed flow: no second gate left), parity 1 failed/11 passed; restore blob == HEAD 46bdaf4b, git diff HEAD empty, preflight --absent clean, both green again (5/5, 12/12). First attempt (bare expression) failed to build, marker never reached dist, readings discarded.",
  "gates": "dispatch-gates --commands on 52b05b8e: 92 derived, 92 run, --ran reconciled (88 run, 4 NOT-MEASURED). check-empty-changeset exit 1 by design (DELIBERATE CORRECTION of pending .changeset/28893-* and 29059-*). NOT MEASURED exit 3 prerequisite: check:dual-build-cjs-loads, check:skill-examples (unbuilt client-react etc.), check:i18n-coverage (showcase needs unbuilt connector-slack dist), check-plugin-teardown-shape --self-test (shallow clone). All others exit 0. Offline: check-changeset-no-major --event exit 0 (no (narrowing) + none, patch), check-adr-0087-registration exit 0. CI check-runs: in_progress at report time.",
  "dogfood_battery": "52b05b8e · VERDICT command-exit 1 · Test Files 2 failed/393 passed/3 skipped (398) · Tests 4 failed/3255 passed/28 skipped (3287) · failures: flow-runas.dogfood.test.ts x3 (runAs system ELEVATES 404 RECORD_NOT_FOUND runas_note; user DE-ELEVATES; system READS), declarative-endpoint-anonymous-guest.dogfood.test.ts x1 (flowContexts length 0). Control: lint/cli src+cli test restored to origin/main 7b1ae338 and rebuilt -> same 4 failed/8 passed. Not this PR's.",
  "mcp_calls": "0",
  "api_writes": "2 — relay pr_create (POST /repos/objectstack-ai/objectstack/dispatches, run 41481992103, PR #29495, 10783 bytes sent = stored); this os-dev-report comment via post-stamped. git push is not a REST write.",
  "line_budget": "12 files in packages: +345/-264 (lint +rule 122, rule test ~100, registry/barrel/explanations +54; cli -122 gate, -50 gate test, -~70 command blocks/ledger/parity table); 4 docs one-line count bumps; 3 changesets (1 new, 2 corrected).",
  "files_changed": [
    ".changeset/29049-flow-region-shape-lint-rule.md",
    ".changeset/28893-flow-region-shape-at-validate-build.md",
    ".changeset/29059-flow-region-gate-hardening.md",
    "packages/lint/src/validate-flow-region-shape.ts",
    "packages/lint/src/validate-flow-region-shape.test.ts",
    "packages/lint/src/authoring-rules.ts",
    "packages/lint/src/index.ts",
    "packages/lint/src/rule-explanations.ts",
    "packages/cli/src/commands/validate.ts",
    "packages/cli/src/commands/compile.ts",
    "packages/cli/src/utils/flow-region-shape.ts (deleted)",
    "packages/cli/src/utils/flow-region-shape.test.ts (deleted)",
    "packages/cli/test/authoring-rule-command-parity.test.ts",
    "packages/cli/test/validate-build-gate-parity.test.ts",
    "packages/cli/test/validate-flow-region-shape.test.ts",
    "content/docs/deployment/cli.mdx",
    "content/docs/deployment/validating-metadata.mdx",
    "content/docs/getting-started/build-with-claude-code.mdx",
    "content/docs/ui/react-pages.mdx"
  ],
  "deviations": [
    "PR body line 维护者速读/需要你做的 still says the rule id goes to the maintainer with needs-user-decision; written before the seat's correction. Corrected body r2 (also carries the dogfood reading) at /mnt/project-files/spec-seat4/29495-body-r2.md for the seat to send; dev does not PATCH.",
    "Session-context asked to set PR assignee/reviewer via MCP; skipped: the standard order forbids MCP writes and os-dev forbids PR assignees.",
    "check-empty-changeset red by design: two pending changesets corrected (DELIBERATE CORRECTION); needs the seat's confirmation, not a restore.",
    "os validate/os build --json row shape changes for this refusal (rule finding in errors / issues instead of the unreleased VALIDATION_ERROR row with flow); verdicts and exit codes unchanged."
  ],
  "open_questions": [
    {
      "question": "New public rule id for the region check (plus exports validateFlowRegionShape / FLOW_REGION_SHAPE_INVALID).",
      "options": [
        "A flow-region-shape-invalid (built with it)",
        "B flow-region-malformed",
        "C flow-region-unregisterable"
      ],
      "recommendation": "A: names the subject (flow region shape) and the verdict in the registry's existing -invalid spelling (approval-expression-invalid, expression-invalid)."
    },
    {
      "question": "Under os lint (which never parses) a flow record FlowSchema refuses is reported by this rule as 'could not be judged' (#29243 fail-closed). validate/build already refuse such a flow at the schema step. Keep, or skip under lint?",
      "options": [
        "A keep fail-closed on every command (built)",
        "B skip schema-refused records (would reopen the false-clean branch at validate/build if FlowSchema ever stops being idempotent)"
      ],
      "recommendation": "A: the engine refuses to register such a flow too, and the rule cannot tell which command runs it."
    }
  ],
  "security_findings": "none",
  "out_of_scope_findings": [
    "class: a · reach: maintainer-visible main-red candidate — pnpm --filter @objectstack/dogfood test on origin/main code (7b1ae338 lint/cli source, closure built) fails 4: flow-runas.dogfood.test.ts (runAs system ELEVATES: trigger answers 404 RECORD_NOT_FOUND on runas_note; user DE-ELEVATES; system READS) and declarative-endpoint-anonymous-guest.dogfood.test.ts (flowContexts length 0). Deterministic on re-run. Possibly from #29381 (objectql nested-write refusal, newest main commit) — not bisected. dedupe words: flow-runas dogfood RECORD_NOT_FOUND runas_note; anonymous guest flowContexts",
    "Top-level graph cycle (#29051): measured not covered — start->a->b->a, b->end draws no finding; the same cycle in a loop body draws 'region contains a cycle'. carrier: #29051"
  ]
}

Stamped 2026-11-02T17:09Z.

Generated by Claude Code

domain:spec#4 · os-steve · session_01ADHkmQNap7ryQcyQXocWct

Claim revision · 2026-11-02T17:51Z · claim 6580500987

This replaces the claim's sentence that "the seat takes [the new rule id] to the maintainer as needs-user-decision". There is no such rule: a new public name is the seat's call unless it carries a real product or contract fork. The seat ruled the id flow-region-shape-invalid in ACCEPT #29495 (comment), and no needs-user-decision label was ever set. The file surface is unchanged: the PR also corrected the pending .changeset/28893-* and .changeset/29059-*, inside the claimed root .changeset/.

domain:spec#4 · os-steve · session_01ADHkmQjqkXDjc5jhjFfHt3
os-dev-report

{
  "issue": 29049,
  "status": "done",
  "branch": "claude/project-thread-f473q9",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/29495",
  "session": "session_01ADHkmQjqkXDjc5jhjFfHt3",
  "premise_still_valid": true,
  "summary": "Addendum (round 2): merged origin/main c88467d4 into the branch, head b3d46b7b. The only conflict was packages/lint/src/authoring-rules.ts, and both sides were append-only: I kept this card's validateFlowRegionShape entry and main's lintConnectorCredentialLiterals entry and their imports. Rule total recounted from the rebuilt lint dist: validate 62, build 62 (lint 60). The four CLI transcripts went from 61 to 62. No other count pins. Delta vs origin/main is unchanged: 19 files, +371/-270. Nothing else touched.",
  "tests": "Head b3d46b7b, after pnpm install --frozen-lockfile and a rebuild of the dogfood and cli closures (VERDICT exit 0). lint test: 192/192 files, 7884/7884 tests, including validate-flow-region-shape. cli unit: 374/374 files, 5271/5271 tests, including validate-flow-region-shape, explain-rule-id, validate-build-gate-parity and the published-subpath pins. cli integration authoring-rule-command-parity: 12/12. check:docs-transcript-drift exit 0 (4 declared transcript values equal the registry's).",
  "gates": "Round-1 gate set on 52b05b8e stands. This round touched only authoring-rules.ts (merge resolution) and the four transcript counts. check:docs-transcript-drift was re-run on b3d46b7b: exit 0. CI should now run, since the merge clears mergeable_state dirty.",
  "dogfood_battery": "b3d46b7b · VERDICT command-exit 0 · Test Files 397 passed / 3 skipped (400) · Tests 3277 passed / 28 skipped (3305) · no failures. The 4 round-1 failures (anchor #29426) are gone with main's fix merged.",
  "mcp_calls": "0",
  "api_writes": "1 — this addendum comment via post-stamped. git push is not a REST write.",
  "line_budget": "unchanged from round 1 apart from the merge: delta vs origin/main is 19 files, +371/-270",
  "files_changed": [
    "packages/lint/src/authoring-rules.ts (merge resolution)",
    "content/docs/deployment/cli.mdx",
    "content/docs/deployment/validating-metadata.mdx",
    "content/docs/getting-started/build-with-claude-code.mdx",
    "content/docs/ui/react-pages.mdx"
  ],
  "deviations": [],
  "open_questions": [],
  "security_findings": "none",
  "out_of_scope_findings": [
    "Round-1 dogfood main-red candidate withdrawn: per the seat it is anchor #29426, fixed by #29451. Dogfood is green on b3d46b7b."
  ]
}

Stamped 2026-11-02T18:03Z.

Generated by Claude Code

domain:spec#4 · os-steve · session_01ADHkmQNap7ryQcyQXocWct

Landed · 2026-11-02T18:44Z

PR #29495 merged as 7f368f56ec at 2026-11-02T18:43:42Z. It landed at head b3d46b7b, after contract review PASS 6583593915 on 52b05b8e, a delta PASS on the merge head (6584199633) and the seat's ACCEPT 6584225221. os lint, os validate and os build now refuse a flow whose loop, parallel or try_catch region the engine refuses to register, through the gating rule flow-region-shape-invalid. The CLI-only walk is retired. pm:dispatched removed in this stroke.

Carried: review should-fix 2 to #29233 (6583607752), which asks it to wire the publish door through this registry row and retire its surfaceReason. The top-level graph cycle stays with #29051.

Navigation Menu

Repository navigation

lint: os lint stays blind to a flow region the engine refuses to register; validate and build refuse it since #28893 through a CLI-only gate #29049

What happens

Wanted

Unblocked · 2026-11-02T07:24Z

Triage: first-touch audit · p3 kept · `domain:devx` → `domain:spec` · `pm:queue` · 2026-11-02T14:15Z

Landed · 2026-11-02T18:44Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

lint: os lint stays blind to a flow region the engine refuses to register; validate and build refuse it since #28893 through a CLI-only gate #29049

Description

What happens

Wanted

Activity

objectstack-fleet commented on Nov 2, 2026

Unblocked · 2026-11-02T07:24Z

objectstack-fleet commented on Nov 2, 2026

Triage: first-touch audit · p3 kept · domain:devx → domain:spec · pm:queue · 2026-11-02T14:15Z

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

Landed · 2026-11-02T18:44Z

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first-touch audit · p3 kept · `domain:devx` → `domain:spec` · `pm:queue` · 2026-11-02T14:15Z