audit-hygiene

name	audit-hygiene
description	Walk the full ADEPT Code Hygiene Quick Reference Card checklist and report PASS/FAIL/N-A for each item. Use as a final check before pushing.
disable-model-invocation	true
allowed-tools	Bash(git *) Read Glob Grep

You are assisting an ADEPT developer with a final code hygiene audit.

Do not use emojis in any output. Enforce all rules in docs/development/AGENT_CODING_RULES.md throughout this skill execution.

Step 1: Read the full code hygiene guide

Read docs/development/CODE_HYGIENE.md in full, paying special attention to the Quick Reference Card (the checklist near the end of the file).

Step 2: Gather state

Run:

git branch --show-current
git status
git log -1 --format="%H %s"
git log -1 --format="%B"
git remote -v
git log --oneline origin/$(git branch --show-current)..HEAD 2>/dev/null | wc -l

Step 3: Walk the checklist

Evaluate every item in the Quick Reference Card. Report each as PASS, FAIL, or N-A with a brief justification.

BRANCH

• Branch naming: Does the current branch name follow <type>-<workstream>-<FY-quarter>-<scope> convention?
• Correct parent: Was the branch created from the correct parent? (Check git log --oneline --graph -10 for merge base.)
• Pushed to origin: Are there unpushed commits to origin? (Check git log --oneline origin/<branch>..HEAD. Developers with personal forks should also check their fork remote.)

IMPLEMENTATION

• Session doc created: Does a SESSION_<N>_*.md file exist in docs/implementation-reports/ or docs/bugfixes/ for the current session?
• Root cause analysis: For bugfixes, does the session doc contain a "Root Cause Analysis" section?
• Code references: Does the session doc use file:line notation for code references?

TESTING

• Timestamped logs: Do test log files exist in logs/ with timestamps from this session?
• All relevant tiers executed: Based on what files changed, were all required tiers run? (Use the same logic as /validate-tests.)
• Test counts recorded: Do the session doc and/or commit message include specific test counts (N passed, M failed)?
• Test claims validated: For every test-outcome claim in docs (session report, validation report, KNOWN_ISSUES, CHANGELOG), verify the claim against an actual timestamped log in logs/. If a doc says "N/N passed" or "failure resolved," the log must exist and show that result. Flag any unvalidated claims.
• Pre-existing failures documented: If any test failures exist, are they explicitly documented as pre-existing?
• Services rebuilt: If files under src/ changed, verify that affected service containers have been rebuilt (image newer than source changes). Use the source-to-service mapping from /validate-tests to identify which containers need checking. Compare docker inspect --format='{{.Created}}' <container> against source file timestamps.
• External stacks re-registered: If MCP tool signatures changed, is there evidence of re-registration?

COMMIT

• Conventional Commit format: Does the most recent commit message follow <type>(<scope>): <description>?
• Body has what+why+test evidence: Does the commit body explain what changed, why, and include test counts?
• Related documentation (this commit): Is this footer section present in the commit message?
• Related documentation (past 5 commits): Is this footer section present with commit hashes?
• No AI agent references: Does the commit message contain zero references to Claude, Gemini, ChatGPT, Copilot, or any AI agent in Author, Co-Authored-By, Signed-off-by, or body text?

DOCUMENTATION

• CHANGELOG.md updated: Does docs/CHANGELOG.md contain an entry for the current session with GitHub Issues: line?
• ROADMAP.md updated: Does the Recently Completed section of docs/ROADMAP.md have an entry for this session's work with GitHub Issues: line?
• KNOWN_ISSUES.md updated: If issues were discovered or resolved, is docs/KNOWN_ISSUES.md updated?
• Sprint tracker updated: If any GitHub issue states changed, is the active sprint status tracker (ls -t docs/implementation-sprints/SPRINT_*_STATUS.md | head -1) updated and in parity with GitHub? (Run gh issue view <N> --json state to verify.)
• Per-issue parity: For each issue that changed state, run grep -rn '#N' docs/ --include='*.md'. Does every local reference (sprint tracker, session tracker, ROADMAP, CHANGELOG, KNOWN_ISSUES, implementation plans/reports) show the correct status? Flag any stale references.
• CLAUDE.md updated: If architectural patterns changed, is CLAUDE.md updated?

Step 4: Present the report

Output a formatted checklist:

ADEPT Code Hygiene Audit
========================

BRANCH
  [PASS] Branch naming: feature-platform-optimizations-FY26Q2-...
  [PASS] Correct parent: branched from feature-platform-optimizations-FY26Q2
  [FAIL] Pushed to origin: 3 commits unpushed to origin

IMPLEMENTATION
  [PASS] Session doc: docs/implementation-reports/SESSION_NNN_...
  [N-A]  Root cause analysis: not a bugfix
  [PASS] Code references: file:line notation used

TESTING
  ...

COMMIT
  ...

DOCUMENTATION
  ...

Summary: 18/20 PASS, 1 FAIL, 1 N-A

Step 5: Remediation

For each FAIL item, provide the specific action needed to fix it. Example:

REMEDIATION:
1. [Pushed to origin] Run: git push origin <branch>

If all items PASS, confirm: "All hygiene checks pass. Ready to push."