fix(skills/pr-management-code-review): stop the AI-attribution footer asserting an unconfirmed maintainer role

[AmirF194]

Summary

• pr-management-code-review's COMMENT footer always claimed an
  <PROJECT> maintainer confirmed the review, but COMMENT is the one
  disposition GitHub lets any account post regardless of write access, so a
  non-collaborator's comment could carry a false claim (feat(codex): first-class Codex CLI skill runtime #902).
• APPROVE/REQUEST_CHANGES are unaffected: GitHub itself refuses those
  mutations without write access, so their existing wording is always
  accurate.
• Fix: COMMENT now renders one of two verbatim footers, chosen from the
  collaborator-permission result prerequisites.md § 1 already computes,
  instead of always asserting maintainer confirmation.

Type of change

• Skill change (skills/pr-management-code-review/): eval fixtures updated below
• Tool / bridge contract (tools/<system>/*.md)
• Python package (tools/*/ with pyproject.toml)
• Groovy reference impl
• Cross-cutting (RFC, AGENTS.md, sandbox, privacy-LLM)
• Documentation (docs/, README.md, CONTRIBUTING.md)
• Project template (projects/_template/)
• CI / dev loop (prek, workflows, validators)
• Other:

Test plan

• prek run --all-files passes (not run as a single command; see Notes)
• For Python packages touched: uv run pytest / ruff check / mypy passes
• For Groovy bridges touched: command-line invocation tested end-to-end
• For skill changes: eval suite passes for the affected skill (see Notes)
• For skill behaviour changes: a new eval fixture is included
  (step-7b-review-body-attribution/fixtures/case-4-footer-role-neutral-comment)
• Other: uv run --project tools/skill-and-tool-validator --group dev skill-and-tool-validate (0 hard failures repo-wide), npx markdownlint-cli2 and typos on all 8 changed files (clean), lychee link check (0 broken links), tools/dev/add-license-headers.py on all 8 changed files (no changes needed)

RFC-AI-0004 compliance

• HITL: any new mutation is gated on explicit user confirmation
• Sandbox: no new unrestricted host access; network reach declared in the adapter
• Vendor neutrality: placeholders (<PROJECT>, <tracker>, <upstream>, <security-list>) used in all skill / tool prose (the check-placeholders prek hook is the mechanical gate)
• Conversational + correctable: agentic-override path documented if behaviour is adopter-tunable
• Write-access discipline: no autonomous outbound messages; drafts only, sent on confirmation
• Privacy LLM: private content does not reach a non-approved LLM; redactor invoked where needed

Linked issues

Fixes #923

Notes for reviewers

I ran the individual hooks the change touches (license headers, markdownlint,
typos, lychee, skill-and-tool-validator) rather than the full prek run --all-files, since installing prek itself was not practical in this
environment; I did not find a diff against any of these tools' output. The
eval runner (skill_evals.runner, print mode) confirms the new
step-7b-review-body-attribution case renders a schema-valid prompt against
the updated system-prompt.md, but I do not have a compatible model CLI
here to run the automated --cli judge pass, so the case is unverified
against an actual model response. Also unverified: a live end-to-end post
from a second GitHub identity without write access to this repo.