Post: Quality Checks, Review, and Formatting

Run all post-implementation checks and fixes.

Prerequisites: Must be in a feature worktree directory with code already implemented and tests built.

Step 1: Quality + Review + Review Tests (parallel)

Run /quality skill, /review skill, AND /review-tests skill simultaneously — they are independent checks (quality is code style/types/slop, review is deep code review, review-tests is test quality) and can execute in parallel.

Step 2: Fix (single writer)

The Step 1 skills are analysis-only; they produce findings, not edits. After all three complete, spawn ONE fix agent (opus, high effort) that owns every edit. It is a single writer on purpose: concurrent fixers race and clobber each other's edits.

Instruct the fix agent to:

1. Read all three findings artifacts: every file in quality-results/, plus review.md and test-review.md.
2. Dedupe: multiple skills often flag the same code location or the same problem — keep one entry.
3. Print the consolidated list before fixing.
4. Verify each finding against the actual code before applying it. Peer findings have been wrong — one claimed a regex hole that did not exist, and "fixing" it would have introduced one. Skip findings that don't hold up, and say which.
5. Apply the remaining fixes serially, one at a time. If a fix moves code referenced by a later finding, re-read the affected file. Skip findings already resolved by an earlier fix.
6. Briefly summarize what was fixed and what was rejected.

Step 3: Lint, Format, Test, Ship

After the fix agent is done, run the quality check script (in this skill's own directory — the "Base directory for this skill" path given above, not a hardcoded ~/.claude path) to auto-format and lint:

bash "<skill-base-dir>/quality_check.sh" --fix

Then verify no errors remain:

bash "<skill-base-dir>/quality_check.sh"

Substitute <skill-base-dir> with the literal base-directory path given above. If errors remain, read the files and fix manually, then re-run until clean.

Run the project's tests and make sure they pass.

Commit and push all changes.

Finally, clean up the findings artifacts:

rm -rf quality-results review.md test-review.md