test-audit

Run the audit with the 'all' argument to evaluate every test file in the project against the Q1-Q25 checklist and AP anti-pattern catalog. Files are batched into groups of 8-10, scored by tier, and aggregated into a summary report saved under zuvo/audits/.

It detects 32 anti-patterns including zero-assertion tests, tautological mocks, duplicate test names, hardcoded waits, committed focus markers like it.only, and flake-masking retries. Red flags such as AP13, AP16, and AP31 trigger automatic Tier-D classification.

Yes, the checklist applies across languages, with stack-specific tools for Python, Kotlin, PHP, and frameworks like Django and FastAPI. Language-specific items such as AP25 are marked N/A where pytest, Go, or Rust have no equivalent.

Passing a file path forces deep mode with full evidence for that file, and multiple files can be listed space-separated. E2E tests are excluded by default and only included when the --include-e2e flag is set.

The audit falls back to standard find, Read, Grep, and Glob commands for file discovery and production pairing. Analysis is slower and production file pairing relies on path conventions instead of symbol resolution.

It is not for single-file code review, writing new tests, or fixing anti-patterns across many files; those belong to the review, write-tests, and fix-tests skills respectively. Use --read-only when you want a report without coverage or backlog writes.