renewisepunk
UserMarkdown-native framework for running a fleet of Claude Code agents in parallel — extracted from Paul9, where it shipped 40+ PRs in 8 days with one human directing.
Categories
Indexed Skills (7)
adopt-framework
This skill should be used when an agent or user asks to "integrate this framework", "install the framework", "adopt the agentic engineering framework", "set up the framework in my repo", "onboard this repo to the framework", or "get started with the framework" — especially into an existing repo that already has docs, plans, a CLAUDE.md, its own tools/, or CI. Installs the framework into a target repo non-interactively and migrates existing markdown content into the ai/ structure.
e2e-auth
This skill should be used when an agent or user needs to run or author an AUTHENTICATED browser test or walkthrough — "test this as a logged-in user", "run the e2e / smoke", "verify the dashboard", "the smoke can't log in", "sign in for tests", "set up Playwright auth", "test the authed flow". Covers getting past the auth provider's bot-detection / MFA (Clerk Testing Tokens; Auth0 / NextAuth / Supabase / custom equivalents), caching the session with storageState, and why auth-gated specs use Playwright, not agent-browser.
efficiency-review
Spawn a fresh agent that reviews the current branch through an efficiency lens — query plans, N+1, bundle size, hot-path latency, and Efficiency budget compliance from the plan. Use when the efficiency gate is required in gates.manifest.json, or standalone on any PR that touches DB queries, hot paths, or shipping dependencies.
independent-review
Spawn an independent reviewer agent that re-reviews the current branch with no implementer context. Use after the implementing agent has written its own review.md, or as part of the /ship-feature flow. Catches the class of bugs that single-agent self-review systematically misses.
new-feature
Start a new feature for this project. Use when the user says "start a feature", "plan X", "new feature", "begin work on", "let's build X", or picks up a Linear issue to work on. Sets up the run folder, writes a plan with the Three-Surface Rule and test approach baked in, and claims the Linear issue.
security-review
Spawn a fresh agent that reviews the current branch through a security lens — authentication, authorization (especially cross-tenant), input validation, injection, secrets/PII, rate limiting, dependency vulnerabilities. Use when the security gate is required in gates.manifest.json, or standalone on any PR that touches auth, HTTP routes, or service actions.
ship-feature
Run after a feature is implemented. Classifies which validation gates apply to this diff, runs only those (acceptance specs, user-value walkthrough, security review, efficiency review, eval suites), aggregates findings, writes review.md, runs compound, posts to Linear, and closes the issue. Use when the user says "ship it", "done implementing", "check my work", "run tests", "wrap up this feature", "close the issue", or "mark it done".
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.