Skip to content

List identity, test guidance, and self-review rubric in README's generated skill structure - #167

Merged
dmccoystephenson merged 1 commit into
mainfrom
fix/readme-generated-skill-structure
Oct 24, 2026
Merged

dmccoystephenson merged 1 commit into
mainfrom
fix/readme-generated-skill-structure

Conversation

@dmccoystephenson

Copy link
Copy Markdown
Member

Summary

A Stage A documentation sweep was run: README.md, SECURITY.md, CONTRIBUTING.md, .github/PULL_REQUEST_TEMPLATE.md, .github/ISSUE_TEMPLATE/*.md, and scripts/README.md were each checked against create-dev-loop.md. Two gaps were found in README.md, and both are fixed here.

  • "Generated skill structure" left out three per-repo substitutions. Step 4 fills IDENTITY, TEST_GUIDANCE, and SELF_REVIEW_RUBRIC from the target repo (create-dev-loop.md:620, :637, :640), and the template uses them at :131, :286, and :381–384. The README did not mention any of them. SECURITY.md already names SELF_REVIEW_RUBRIC as content shaped by the target repo, so the README undersold what a generated skill actually encodes. Bullets for all three have been added.
  • Step 2's summary line left out sources Step 2 actually reads. The README listed only part of the Step 2 read list. README.md, formatter configs, documentation sources, the PR template, and recent commits are now included (create-dev-loop.md:58–96), which also matches SECURITY.md's trust-model list.

The other swept docs showed no drift. CLAUDE.md was not edited: it is harness-blocked, and its known drift is tracked in #106.

No tracking issue: the gap was found during triage.

Skipped issues: every other open issue (#106, #109, #114, #120–#125, #127, #129–#131, #133, #135, #140–#150, #158, #165) was deferred. Each one either changes template behavior, which cannot be validated headlessly per #109 and so could not merge this run, or requires editing CLAUDE.md (#106). Per RESEARCH.md §2's tiebreaker, a documentation fix was preferred. One finding was noted: #123 and #142 are partly addressed already by the "Standing merge authorization does not satisfy the anchor gate" paragraph (create-dev-loop.md:498, from PR #119). That paragraph says a flag passed before the PR existed does not satisfy the codeowner exception, but it is framed around the UNVERIFIED case and has no Edge cases entry. The issues are left open for that remaining work.

Doc sync check

  • README.md's "What it does" Step list still matches create-dev-loop.md's Steps 1:1 (no Step was added or removed; check_docs.py passes)

Test plan

  • python3 scripts/check_docs.py → Doc consistency check passed.
  • python3 -m unittest discover -s tests -v → Ran 41 tests ... OK
  • Only docs changed: create-dev-loop.md was not touched, so no /create-dev-loop behavioral run applies (UNVERIFIED-not-applicable).

This PR description was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).

🤖 Generated with Claude Code


drafted by Claude on behalf of Daniel Stephenson

…rated skill structure

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@dmccoystephenson

Copy link
Copy Markdown
Member Author

Self-review rubric (anchored on CI run 40049988646, doc-consistency pass):

  • Scope: PASS — README.md is the only file modified (+4/−1), and every hunk serves the README-versus-Step-2/Step-4 drift fix.
  • Tests-new: PASS (n/a) — no code was changed.
  • Tests-fix: PASS (n/a) — this is a docs correction, not a code bug fix, so there is nothing to revert and re-run.
  • Sibling structure: PASS — no new file was added. The new bullets follow the existing - **Name** — description form.
  • Sibling renames: PASS (n/a) — nothing was renamed.
  • Docs: PASS — the new Identity bullet was checked against create-dev-loop.md:620, which says to infer it from "CLAUDE.md, CONTRIBUTING.md, and 2-3 recent PRs". Test guidance was checked against :637 ("Framework name, naming convention, file location, any fake/mock patterns"). Self-review rubric was checked against :640 ("answerable from the diff or a command output … Omit the conditional block entirely if no repo-specific items apply"). The Step 2 additions were checked against :62–96. SECURITY.md's trust-model list already includes the same sources, so no further sync is needed.
  • Issue resolution: PASS (n/a) — there is no Closes #N. The gap was found during triage.
  • CI: PASS — doc-consistency is green on the PR head. Locally, check_docs.py passed and Ran 41 tests ... OK.
  • Placeholder parity / Fence escaping / Phase numbering / Version comments / No back-ported specifics / gardener interface / Retrofit note: PASS (n/a) — create-dev-loop.md is untouched.
  • README Step parity: PASS — no Step was added or removed. Only the wording of Step 2's bullet changed, and check_docs.py confirms the list is still 1:1.
  • Research grounding: PASS — the PR is docs-only. The PR body cites RESEARCH.md §2 for choosing docs work, and no finding is shipped.
  • Behavioral validation: UNVERIFIED-not-applicable — the template was not changed.

Judgment note: README.md:72 — the Identity bullet leaves out the "repo README pitch" source named in Step 2's question table (create-dev-loop.md:115). It follows the narrower Step 4 row instead. This is left as is, because the Step 4 row is the contract for substitution.

Summary: this is a docs-only drift fix. CI is green and no blocking findings were raised.

This review comment was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).


drafted by Claude on behalf of Daniel Stephenson

@dmccoystephenson
dmccoystephenson merged commit 8ce1f40 into main Oct 24, 2026
1 check passed
@dmccoystephenson
dmccoystephenson deleted the fix/readme-generated-skill-structure branch October 24, 2026 01:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant