Loading...
Loading...
Found 107 Skills
TDD feature build loop: spec (RED) → implement (GREEN) → refactor. Pass the feature name as argument.
Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation
Use this when you are about to start Discover for an existing project, but you don't know "which portals can be used as evidence (run/test/ci/contract/ops)" and "which modules should be prioritized (P0/P1/P2)", and you are worried that out-of-control scope will lead to unmaintainable results.
Make all tech decisions, write CLAUDE.md, scaffold the project, and get a smoke test passing. Run after /wireframe.
Generate clickable HTML wireframes for every screen in the PRD. No design tool needed — open in browser.
8-agent QA loop: browser exploration via Playwright MCP, then analyze, plan, test, audit, heal, expand, snapshot. Quality gate score >= 85 to pass.
Used to confirm that development is completed and ensure all tests/checks have passed
Start Task
Implement a project from its documentation and specification. Use when asked to "implement project", "continue implementation", "build from docs", "implement from spec", or when the user wants to progressively implement a documented project following a todo checklist. Reads docs/, creates implementation plans and todo lists, and implements incrementally with tests and commits.
Use when asked to fix, resolve, or address a GitHub PR — including review comments, Copilot suggestions, and CI test failures. Triggers on phrases like "fix PR", "fix PR comments", "resolve review", "address Copilot feedback", "fix review comments", "fix CI", "fix failing tests on PR", or when given a PR number to fix.
Use when you need to produce a D2 decision document (RFC/Decision Doc) for a Spec Pack, and need to independently decide whether to execute D0 (bypass design via diversion) and D1 (optional research) even under pressure, avoiding guessing FEATURE_DIR, making up input details, writing design as implementation specifics, or leaving behind TODO/To-Confirm lists.
Use when you need to perform D1 research during the Spec-level design phase (output `{FEATURE_DIR}/design/research.md`), or when you face critical uncertainties/high-risk points that require verification first instead of directly proceeding to D2; common symptoms include lack of evidence to support trade-offs, unknown items written as TODOs/to-be-confirmed issues, attempting to guess the FEATURE_DIR under pressure, or writing research as implementation details.