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
@microsoft
openclaw
@openclaw
OpenAI
@openai
The Apache Software Foundation
@apache
GitHub
@github
Anthropic
@anthropics
AutoGPT
@Significant-Gravitas
Garry Tan
@garrytan全部 Skills 192262
openspec-sync-specs
Sync delta specs from a change to main specs. Use when the user wants to update main specs with changes from a delta spec, without archiving the change.
openspec-propose
Propose a new change with all artifacts generated in one step. Use when the user wants to quickly describe what they want to build and get a complete proposal with design, specs, and tasks ready for implementation.
review-skills
Audit all skills available in the current project context for conflicts, quality issues, and replicable patterns. Use when user wants to review skills, audit skills, check skill quality, find conflicting or overlapping skills, improve skill descriptions, check what skills are installed, or mentions "review-skills". Always use this skill when the user wants to understand the health or completeness of their skill setup.
openspec-onboard
Guided onboarding for OpenSpec - walk through a complete workflow cycle with narration and real codebase work.
openspec-new-change
Start a new OpenSpec change using the experimental artifact workflow. Use when the user wants to create a new feature, fix, or modification with a structured step-by-step approach.
repo-skill-candidates
Audits one codebase for the recurring, multi-step patterns worth turning into repo-level Claude Code skills, and writes a short evidence-backed report to skill-candidates-report.md. Evidence comes only from the code (source, layout, tests, CI, scripts, docs, existing .claude setup), so it works on a new or inherited repo with no history. Triages every pattern to skill, CLAUDE.md line, linter config or script, and drops what existing skills already cover. Use when the user says "find skill candidates", "what in this repo should be a skill", "audit this repo for skill opportunities", "which patterns here are worth capturing as skills", or "repo-skill-candidates". Report only. Not for writing or drafting a skill (use skill-creator), reviewing code quality (use code-review or architecture-review), or suggesting skills from session history or the user's own habits (use reflect).
openspec-ff-change
Fast-forward through OpenSpec artifact creation. Use when the user wants to quickly create all artifacts needed for implementation without stepping through each one individually.
openspec-explore
Enter explore mode - a thinking partner for exploring ideas, investigating problems, and clarifying requirements. Use when the user wants to think through something before or during a change.
reflect
Review a session's transcripts (the main agent and every subagent it spawned) for confusion and waste, such as hallucinated paths and symbols, repeated failures, edit churn, user corrections, expensive or redundant tool calls, long hunts for a file and review misses; audit each repo it touched for missing guardrails and instruction bloat; then file GitHub issues for the fixes the user approves. Also reports instruction-file metrics. Use only when invoked as /reflect, /reflect <session id>, /reflect metrics or /reflect status.
openspec-continue-change
Continue working on an OpenSpec change by creating the next artifact. Use when the user wants to progress their change, create the next artifact, or continue their workflow.
openspec-bulk-archive-change
Archive multiple completed changes at once. Use when archiving several parallel changes.
find-skills
Not for building skills — for finding them. Use when the intent is to discover, search, or inventory what agent skills already exist for a topic. The user wants to know "what's out there for X" rather than "make me something." Triggers on: "find a skill for X", "skill search: X", "what skills exist for X", "discover skills for X", "is there already a skill for X", "before I build this, has someone done it?", or any pre-build audit. Also trigger proactively before any skill-creator session when the user hasn't searched yet — finding what exists prevents duplicate effort. Output is a ranked table plus a thematic synthesis document, never a new skill.
openspec-archive-change
Archive a completed change in the experimental workflow. Use when the user wants to finalize and archive a change after implementation is complete.
openspec-apply-change
Implement tasks from an OpenSpec change. Use when the user wants to start implementing, continue implementation, or work through tasks.
update-project-docs
Sync documentation with the codebase and generate token-lean architecture codemaps. Use when user wants to update docs, sync documentation, generate codemaps, update architecture docs, run /update-docs, or run /update-codemaps.
testing-policy-creator
Writes, or brings up to date, a repo's own testing-policy skill at .claude/skills/testing-policy/SKILL.md: which code gets which kind of test, with which doubles, asserting what, and why. Surveys the codebase for its layers and boundaries, proposes a draft map, grills the user on it row by row, then lists the tests and boundaries that break the policy. Use when the user says "create a testing policy", "write our testing-policy skill", "decide what we test in this repo", "which tests should we keep", "our tests are a mess, set some rules", "purge tests that don't conform", or asks to update an existing testing policy. Not for writing a test (use the language's testing skill and tdd) or reviewing a diff.
tdd
Use when implementing any feature or fixing any bug. TDD is the default — not an optional approach. Trigger on any implementation work, not just when the user explicitly mentions TDD.
tailwind-shadcn
shadcn/ui + Tailwind CSS standards. Auto-detects the framework — React uses the shadcn CLI with Base UI primitives, SolidJS uses shadcn-solid with Kobalte. TRIGGER — load BEFORE writing or editing any styling, not after: a `class=` or `className=` attribute is being added or changed; tailwind.config.*, globals.css, or anything under components/ui/ is touched; the prompt names Tailwind, shadcn, a variant, a design token, a CSS variable, a responsive breakpoint, dark mode, or a colour; or an arbitrary bracket value like [24px] or [#1a1a1a] is about to be written. Trigger on plain requests — "style this", "add a colour", "make this responsive", "match the other buttons" — the user will not say "Tailwind discipline"; the class attribute is the trigger. SKIP only when no styling and no component file is touched.