name: quality description: Run all quality gates and report detailed status. Use as final verification after parallel sub-agents, or as all-in-one check.

All Quality Gates

When to Use

This skill serves TWO purposes:

1. All-in-one check (Codex, Gemini, or simple tasks): Run all quality gates sequentially in a single invocation.

2. Final verification (Claude with parallel agents): After spawning /check, /test, /doc in parallel, run /quality to verify nothing was missed due to concurrent edits.

Either approach works. Don't do both — that's redundant. For Claude on complex tasks: spawn parallel sub-agents → then /quality as final check. For simpler tasks or other agents: just run /quality directly.

Procedure

1. Run all quality gate commands sequentially:

   make check-fix
   make check
   make check-strict-all
   make test-with-coverage
   make doc
   

2. If any fail, fix the issues and re-run from step 1

3. Report detailed status summary:

   ## Quality Gates Summary
   
   ### Linting & Type-Checking
   - `make check`: PASS/FAIL
   - `make check-strict-all`: PASS/FAIL
   - Errors fixed: <count> (list briefly if any)
   - Remaining warnings: <count or "none">
   
   ### Tests
   - `make test-with-coverage`: PASS/FAIL
   - Tests: <passed>/<total>
   - Coverage (overall): <X>%
   - Coverage (new code): <X>% [list uncovered files/lines if < 100%]
   - Integration tests (`make test-all`): PASS/FAIL/SKIPPED
   
   ### Documentation
   - `make doc`: PASS/FAIL
   - New terms added to project-words.txt: <list or "none">
   - New modules documented: <list or "none">
   
   ### Overall: PASS/FAIL
   
   ### Lessons Learned
   <Patterns discovered while fixing issues that should be documented>
   <If none, write "None">
   

Lessons Learned

Always include a "Lessons Learned" section, even if empty. This is critical when running as a sub-agent — the main agent needs this information for the PR description.

Examples of lessons learned (things that should go in AGENTS.md or dev/*.md):

• "Discovered that mypy requires explicit type annotations for class variables"
• "Tests using freezegun must import it before the module under test"
• "The --strict flag requires Optional[] for all nullable parameters"

If you fixed issues, ask: "Would knowing this earlier have helped?" Note it in the lessons learned output. The main agent will include it in the PR description under "Lessons Learned", and the human maintainer will decide whether to update AGENTS.md or dev/*.md.

Exit Conditions

• Success: All gates pass
• Failure: Report which gates failed with specific errors/uncovered lines