GITHUB AGENT SKILLS / 技能收录
给你的 AI 助手
装上专业技能
我们从 GitHub 持续收录、审核并分类优质的 Agent Skills,按职业与使用场景整理,复制一行命令即可安装到 Claude Code、Cursor 等智能体。
按职业浏览 Agent Skills
为不同岗位挑选的高频技能
按使用场景浏览 Agent Skills
从你要完成的任务出发
按创作者浏览 Agent Skills
按 GitHub Stars 排序的优秀创作者
Microsoft
@microsoftopenclaw
@openclawAnthropic
@anthropicsOpenAI
@openaiThe Apache Software Foundation
@apacheGitHub
@githubAutoGPT
@Significant-GravitasGarry Tan
@garrytan全部 Skills 482556
autopilot-intake
Structured one-question-at-a-time intake for Auto Pilot. Use when a new autopilot project starts and the required setup information has not been collected yet.
autopilot-designer
Internal Auto Pilot designer role. Use only from the manager workflow for user-facing projects.
autopilot-architect
Internal Auto Pilot architect role. Use only from the manager workflow when structure decisions are non-trivial.
auto-pilot
Intake-first autonomous project execution for Codex. Use when the user wants a short project request turned into a long-running build workflow with minimal interruptions, or explicitly mentions 오토파일럿/autopilot.
sync-tasks
Warm the task graph and vector store by synchronizing a configured task board through the reviewer MCP server. Use when the user asks to sync or index board tasks.
sync-codebase
Build or update the reviewer base index (vector store + code graph) from a local repo clone. Use when the user asks to "index the codebase", "build the index", "sync the code", "rebuild the graph", "просиндексируй код", "построй индекс".
summarize-subsystems
Precompute concise per-subsystem summaries (GraphRAG community summaries) over the base code index, so ask / PR-walkthrough get a cheap high-level prior. Use when the user asks to build/refresh subsystem summaries ("просуммируй подсистемы", "построй обзоры модулей", "summarize subsystems"). Requires a built base index + the reviewer MCP server.
solve-task
Gather disciplined context for solving a task, then hand off to development. Use when the user asks to solve/implement a task ("solve PRI-4", "rag-reviewer:solve-task <key or description>", "реши задачу X"). Reads a keyed task from the reviewer store, pulls related/similar tasks and relevant code, distills a brief, and enters brainstorming. Requires the reviewer MCP server.
review-pr
Review a GitHub pull request with the RAG + code-graph pipeline (reviewer MCP server). Use when the user asks to review a PR ("review PR 123", "заревьюй PR", a PR URL). Requires ParadeDB/Neo4j running and a built base index.
report-bug
Report a defect of reviewer or its plugin as an anonymized GitHub issue in mimfort/rag_for_git, after explicit user approval. Use when a reviewer tool breaks its own contract, a skill prescribes an impossible step, a stated invariant fails, or the user asks to report a reviewer bug.
pr-walkthrough
Build a human-facing reading guide for a GitHub pull request (where to start, what each file changes, what it impacts) using the reviewer PR session + code graph. Use when the user asks to walk a human reviewer through a PR ("PR walkthrough", "гид по PR", "как читать этот PR", "проведи по PR"). NOT a bug review (see review-pr). Requires the reviewer MCP server + base index.
performance-review
Review code changes only for performance and efficiency risks (N+1 queries, repeated work, bad asymptotics, missing batching/caching, blocking I/O, memory growth). Use when the user explicitly asks for a performance review of a diff/PR.
maintainability-review
Review code changes only for maintainability risks: unnecessary complexity, poor readability, duplication, weak separation of concerns, and misalignment with repository conventions. Use when the user explicitly asks for maintainability review, code quality review, clean code review, complexity review, readability review, simplification review, or how to simplify changed code without changing behavior.
finish-task
After a task's PR is created, offer to close the task on the configured board and reindex it. Use when the user says a PR is up or asks to close or finish a task.
decompose-task
Use when a configured board task needs decomposition into native child tasks.
create-task
Create a task on a configured board with canonical server-side structure. Use when the user asks to file or create a task on the board.
configure-review
Use when configuring or changing a repository's tracked branches, layered review policy, ignored tracked paths, retrieval limits, summary depth, or non-secret task-board metadata.
ask
Answer grounded questions about the codebase (onboarding / Q&A) using the reviewer RAG + code graph. Use when the user asks how the code works, where something lives, or to explain a subsystem ("where is auth", "how does X work", "explain the indexing flow", "как устроено…", "где у нас…", "объясни код"). Requires a built base index + graph (reviewer MCP server). Not for reviewing PRs.