浏览 shami-ah 维护的已发布 AI 技能，比较用途与说明，并通过来源仓库了解安装方式。
已收录技能
32收录技能的仓库 Stars 合计
0仓库 Forks
32平均每个技能的仓库 Stars
0同一仓库中的多个技能可能共享 Stars，合计包含重复仓库数据。
Repo navigation and “where do I start?” guide for OpenEvent-AI. Use when an agent/dev is lost, needs an overview of the project or a subsystem, or needs pointers to the most important docs, entrypoints, scripts...
Review and testing checklist for OpenEvent email workflow changes (intake, offers, availability, confirmation).
Debug “backend won’t start / first request 500 / ports stuck” quickly. Use when uvicorn reload behaves strangely, ports 8000/3000 are blocked, or you see first-request errors like “unexpected keyword argument”...
Maintain and extend prompt-injection defenses. Use when adding new user-input surfaces, changing prompt templates, or when a new injection pattern is observed; run the security regression suite and add a minima...
Setup and verification for MCP tooling used in OpenEvent-AI (filesystem, Context7 docs, LSP/Pyright, Playwright). Use when onboarding a new developer/agent or when Codex/Claude cannot access LSP, browser automa...
Official Playwright E2E testing protocol. Triggers on "E2E", "Playwright", "browser test". Enforces fresh client state, hybrid mode, and clean ports.
Find and replace dangerous keyword-based detection with LLM signals. Use when debugging detection interference or before touching routing/detection code. Produces a prioritized migration plan.
Full E2E verification with REAL APIs (OpenAI + Gemini). Use when stub tests pass but you need to verify real LLM behavior. Requires hybrid mode (I:ope/E:gem/V:ope). NEVER uses AGENT_MODE=stub.
默认按已收录数据中的 GitHub 仓库 Stars 排序。Stars 属于来源仓库，同一仓库的多个技能可能共享相同数值；它反映仓库关注度，不代表单个技能的安装量或质量评分。