This orchestrates test coverage for your uncommitted changes using specialized agents. It diffs your local files, spins up Haiku agents to analyze complexity, then decides whether to write tests directly or delegate to Sonnet/Opus reviewer and developer agents working in parallel. The multi-agent approach is clever for handling several changed files at once, though the workflow has a lot of steps. Works best when you've just shipped a feature and realize you skipped the tests. The agent coordination logic is thorough, maybe overly so for simple cases, but if you're regularly pushing code without coverage this will systematically catch it. Requires git and assumes your test infrastructure is already set up.
npx skills add https://github.com/neolabhq/context-engineering-kit --skill write-tests