docs(B-0171.3): target retraction-native spec.md#6208
Conversation
Why: - B-0171.3 has a narrow row-level correction available after #6207 clarified that strict OpenSpec inventory reads only `spec.md` files. - The claim needs to be visible before editing the row. Scope: - docs/backlog/P1/B-0171.3-author-retraction-native-spec.md - docs/claims/codex-loop-b0171-3-spec-path-20260531.md Agency-Signature-Version: 1 Agent: Vera Agent-Runtime: OpenAI Codex desktop heartbeat loop Agent-Model: GPT-5 Credential-Identity: aaron-codex-desktop Credential-Mode: shared Human-Review: none Human-Review-Evidence: none Action-Mode: autonomous-fail-open Task: B-0171.3 Co-Authored-By: Codex <noreply@openai.com>
Why: - B-0171.3 still pointed acceptance at `openspec/specs/retraction-native/README.md`. - The current OpenSpec inventory gate discovers capability specs only at `<capability>/spec.md`, so the child row should not direct implementation to a README-only artifact. What: - Update B-0171.3 acceptance criteria to require `openspec/specs/retraction-native/spec.md`. - Note that the existing README can remain background material but is not inventory-discovered. - Release the claim file in the same PR branch. Proof: - bun tools/openspec/inventory.ts --enforce --fail-on-unmapped-specs - git diff --check - bunx prettier --check docs/backlog/P1/B-0171.3-author-retraction-native-spec.md Agency-Signature-Version: 1 Agent: Vera Agent-Runtime: OpenAI Codex desktop heartbeat loop Agent-Model: GPT-5 Credential-Identity: aaron-codex-desktop Credential-Mode: shared Human-Review: none Human-Review-Evidence: none Action-Mode: autonomous-fail-open Task: B-0171.3 Co-Authored-By: Codex <noreply@openai.com>
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: cf753337ac
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Why: - Codex review correctly noted that adding `openspec/specs/retraction-native/spec.md` without an inventory map entry would make the strict unmapped-spec gate fail. - B-0171.3 acceptance criteria should preserve the inventory invariant for the implementer. What: - Require `retraction-native` to be mapped in `CAPABILITY_MODULE_MAP` or `CAPABILITY_ARTIFACT_MAP` when the spec is authored. - Keep the strict inventory gate command in the acceptance criterion. Proof: - bun tools/openspec/inventory.ts --enforce --fail-on-unmapped-specs - git diff --check - bunx prettier --check docs/backlog/P1/B-0171.3-author-retraction-native-spec.md Agency-Signature-Version: 1 Agent: Vera Agent-Runtime: OpenAI Codex desktop heartbeat loop Agent-Model: GPT-5 Credential-Identity: aaron-codex-desktop Credential-Mode: shared Human-Review: chatgpt-codex-connector review comment Human-Review-Evidence: #6208 (comment) Action-Mode: autonomous-fail-open Task: B-0171.3 Co-Authored-By: Codex <noreply@openai.com>
There was a problem hiding this comment.
Pull request overview
Retargets the B-0171.3 backlog row's acceptance criteria from openspec/specs/retraction-native/README.md to openspec/specs/retraction-native/spec.md so the future spec is actually discovered by the strict OpenSpec inventory gate. Also bumps last_updated and clarifies that the existing README may remain as non-discovered background material.
Changes:
- Update acceptance criteria to target
spec.md(the inventory-discovered path) instead ofREADME.md. - Note that the existing
README.mdis allowed as background material but is not scanned bytools/openspec/inventory.ts. - Bump
last_updatedto 2026-05-31 per backlog frontmatter convention.
* claim: backlog-0171-4 Why: - Codex/Vera has no active claim after #6208 merged. - B-0171 still needs decomposition before implementation, and no active claim overlaps the B-0171 parent or B-0171.4 child-row path set. What: - Claim a single B-0171 decomposition slice for an atomic backlog-row schema OpenSpec child. - Record the scoped paths before write work. Proof: - git fetch --prune origin - git ls-remote --heads origin 'refs/heads/claim/*' - bun tools/backlog/empty-queue-pickup.ts --json --dry-run --repo-root /Users/acehack/.local/share/zeta-codex-loop/Zeta --worktree-root /Users/acehack/.local/share/zeta-codex-loop/Zeta-worktrees Agency-Signature-Version: 1 Agent: Vera Agent-Runtime: OpenAI Codex desktop heartbeat loop Agent-Model: GPT-5 Credential-Identity: aaron-codex-desktop Credential-Mode: shared Human-Review: not-requested Action-Mode: autonomous-fail-open Task: B-0171.4 Co-Authored-By: Codex <noreply@openai.com> * docs(B-0171): decompose backlog-row schema spec child Why: - The B-0171 selector still reports decomposition-first work. - Backlog row schema is explicitly next in the parent sequence after tick-history, but it lacked an atomic OpenSpec authoring child. - The B-0171.3 review showed new spec children need inventory mapping acceptance criteria up front. What: - Add B-0171.4 for an inventory-discovered backlog-row-schema OpenSpec capability. - Require the future spec to map its schema docs and enforcement/generation tools in CAPABILITY_ARTIFACT_MAP. - Record the current B-0171 child sequence and regenerate docs/BACKLOG.md. - Release the git-native claim in this PR branch. Proof: - bun tools/backlog/lint-frontmatter.ts --file docs/backlog/P1/B-0171.4-author-backlog-row-schema-spec.md --strict - bun tools/backlog/generate-index.ts --check - git diff --check - bunx prettier --check docs/backlog/P1/B-0171-openspec-catch-up-canonical-source-of-truth-aaron-2026-05-03.md docs/backlog/P1/B-0171.4-author-backlog-row-schema-spec.md - bunx markdownlint-cli2 docs/backlog/P1/B-0171-openspec-catch-up-canonical-source-of-truth-aaron-2026-05-03.md docs/backlog/P1/B-0171.4-author-backlog-row-schema-spec.md docs/BACKLOG.md Note: - Prettier check against generated docs/BACKLOG.md is intentionally not used; the canonical check for that generated file is tools/backlog/generate-index.ts --check. Agency-Signature-Version: 1 Agent: Vera Agent-Runtime: OpenAI Codex desktop heartbeat loop Agent-Model: GPT-5 Credential-Identity: aaron-codex-desktop Credential-Mode: shared Human-Review: not-requested Action-Mode: autonomous-fail-open Task: B-0171.4 Co-Authored-By: Codex <noreply@openai.com> * docs(B-0171): close tick-history spec dependency Why: - PR #6209 review found B-0171.4 is marked buildable-now while depending on B-0171.2. - B-0171.2 already has `openspec/specs/tick-history/spec.md`, but its backlog row still said `status: open`. - Leaving that row open would make autonomous pickup block B-0171.4 on `dependency B-0171.2 is open`. What: - Mark B-0171.2 closed with `openspec/specs/tick-history/spec.md` as the closing substrate. - Add a resolution note tying the spec to OpenSpec inventory discovery and mapped tick-history artifacts. - Regenerate `docs/BACKLOG.md` so the closed dependency is reflected in the index. Proof: - bun tools/backlog/lint-frontmatter.ts --file docs/backlog/P1/B-0171.2-author-tick-history-spec.md --strict - bun tools/backlog/lint-frontmatter.ts --file docs/backlog/P1/B-0171.4-author-backlog-row-schema-spec.md --strict - bun tools/backlog/generate-index.ts --check - git diff --check - bunx prettier --check docs/backlog/P1/B-0171.2-author-tick-history-spec.md docs/backlog/P1/B-0171.4-author-backlog-row-schema-spec.md docs/backlog/P1/B-0171-openspec-catch-up-canonical-source-of-truth-aaron-2026-05-03.md - bunx markdownlint-cli2 docs/backlog/P1/B-0171.2-author-tick-history-spec.md docs/backlog/P1/B-0171.4-author-backlog-row-schema-spec.md docs/backlog/P1/B-0171-openspec-catch-up-canonical-source-of-truth-aaron-2026-05-03.md docs/BACKLOG.md - bun tools/openspec/inventory.ts --enforce --fail-on-unmapped-specs Agency-Signature-Version: 1 Agent: Vera Agent-Runtime: OpenAI Codex desktop heartbeat loop Agent-Model: GPT-5 Credential-Identity: aaron-codex-desktop Credential-Mode: shared Human-Review: chatgpt-codex-connector review comment Human-Review-Evidence: #6209 (comment) Action-Mode: autonomous-fail-open Task: B-0171.4 Co-Authored-By: Codex <noreply@openai.com> --------- Co-authored-by: Lior <lior@zeta.dev> Co-authored-by: Codex <noreply@openai.com>
Summary
openspec/specs/retraction-native/spec.mdWhy
#6207 clarified that the strict OpenSpec inventory gate only treats directories with
spec.mdas capability specs. B-0171.3 still directed implementation toREADME.md, which would keep the formal retraction-native capability invisible to the inventory scanner.Checks
bun tools/openspec/inventory.ts --enforce --fail-on-unmapped-specsgit diff --checkbunx prettier --check docs/backlog/P1/B-0171.3-author-retraction-native-spec.mdLimits
docs/BACKLOG.mdor Expose observe hierarchy operating readouts #6200.