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
claude-api
Anthropic Claude API patterns for Python and TypeScript. Covers Messages API, streaming, tool use, vision, extended thinking, batches, prompt caching, and Claude Agent SDK. Use when building applications with the Claude API or Anthropic SDKs.
clarify-requirements
Resolve genuine pre-execution ambiguity by scanning repo context, generating 1-6 grounded options, and asking 1-3 focused questions via AskUserQuestion. NOT for mid-task use (injection risk); NOT when the goal is already unambiguous; use to disambiguate vague prompts before any file edits.
claim-verify-gate
Use when an answer, research finding, audit verdict, or status report is about to be treated as true and the next action depends on it. Decomposes the answer into atomic, individually-checkable claims (SAFE, arXiv 2403.18802), runs Chain-of-Verification (arXiv 2309.11495 - draft, plan verification questions, answer them independently, revise), and re-probes each claim with a MECHANICALLY DIFFERENT instrument than produced it (different tool, engine, shell or measured field; re-running the same probe never counts as CONFIRMED). Emits a per-claim CONFIRMED / REFUTED / UNVERIFIED table with the settling experiment, and forces an overclaim to be rewritten to the scope actually measured - an honestly-scoped residual does NOT block. Ships scripts/claim_gate.py, a deterministic 10-rule checker with a built-in both-states self-test. NOT for build/test/lint claims (hooks/claim-guard and verification-before-completion own those), NOT a test runner, NOT a trained critic - no model judge, no new runtime.
citation-mapping
Use when you need to build a citation or reference map for a paper, tracing backward references and forward cited-by works from metadata APIs like Semantic Scholar and OpenAlex
ci-silent-pass-lint
Use before any PR creation or "gates green / ready to merge" claim to detect CI/gate workflows that report GREEN while masking real findings - the recurring "qlty green yet 61 issues + 809 on Codacy" / stale-SHA-gate pattern. Runs a deterministic linter over .github/workflows/*.yml for 11 silent-pass shapes plus an MCP cross-check of actual finding-counts at HEAD vs the reported status.
cheminformatics-rdkit
Use to drive the standalone rdkit MCP server (tandemai-inc/rdkit-mcp-server, 74 tools, MIT) for license-free chemistry - molecular descriptors, fingerprints, similarity, substructure search, and scaffold analysis on SMILES/structures. Covers the stdio registration and two install gotchas. Use when a chemistry question needs RDKit and does not need a Schrodinger license.
cheat-engine-x64dbg-ida-handoff
Use when moving runtime findings between Cheat Engine, x64dbg, and IDA Pro, especially for converting value scans or breakpoints into typed static analysis and durable module-relative anchors.
cheat-engine-workflow
Use when performing Cheat Engine-driven runtime analysis, including value scanning, pointer sanity checks, AOB validation, structure inspection, and trainer-oriented confirmation during reverse-engineering or game runtime workflows.
chatgpt5-web
Direct browser-automation ChatGPT 5 web lane - one skill, three modes via a `mode:` arg. Use when the user explicitly wants the ChatGPT web side (not Codex orchestration): mode:query for a single direct question; mode:discuss to debate/ideate/spar; mode:curate to compare candidates and pick the best. NOT for Codex-side lanes (use codex-consult).
chatgpt-apps
Build, scaffold, refactor, and troubleshoot ChatGPT Apps SDK applications that combine an MCP server and widget UI. Use when Codex needs to design tools, register UI resources, wire the MCP Apps bridge or ChatGPT compatibility APIs, apply Apps SDK metadata/CSP/domain settings, or produce a docs-aligned scaffold. Prefer a docs-first workflow via openai-docs or the OpenAI docs MCP.
caveman
Ultra-compressed communication mode. Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping full technical accuracy. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman.
caption-burn
Use when BURNING captions / subtitles / karaoke INTO a video as hardcoded pixels (hardsub) via libass through ffmpeg, OR soft-muxing them as a selectable, removable subtitle track. Covers ASS document generation sized to the export canvas, libass styling (font / size / colour / outline / box), the OpusClip word-by-word KARAOKE preset (per-word alternating yellow/green + scale-pop), hook-title / hook-card overlays, the CaptionOverride tuning patch, safe-area placement for 9:16, and the Netflix CPS/CPL/min-gap polish pass. Burn = lossy re-encode, permanent, styling survives anywhere; soft-mux = lossless stream-copy, toggleable, but loses ASS styling in MP4 (mov_text). NOT for moving/animated watermark removal (use video-watermark-removal), NOT for transcription/cue generation (this consumes an existing cue list), NOT for repositioning an already-burned track (burned pixels are permanent — re-burn from source). Source of truth: media_studio/features/caption.py, caption_karaoke.py, caption_override.py, caption_polish.py, tracks.py.
c4-architecture
Use when asked to create architecture diagrams, document system architecture, visualize software structure, or generate context/container/component/deployment diagrams — produces C4 model Mermaid diagrams written to docs/architecture/.
browser-rail
Use as the DEFAULT for any multi-step browser work on a site the owner is signed into - reading, filling, clicking, uploading, or posting across several turns on grok.com, chatgpt.com, a dashboard, a webmail, an admin panel, or any logged-in app. Drives a dedicated CDP Chrome/Edge profile with Playwright over role locators (the a11y tree), not pixels. Covers the Chrome-136 break where --remote-debugging-port is silently IGNORED on the default user-data-dir, the dedicated-profile fix, how the owner signs in ONCE so no credential ever passes through an agent, passing content as an evaluate ARG instead of inlining it into JS, and when to fall back to claude-in-chrome instead. Triggers include drive the browser, log into a site and do X, automate this web app, upload a file to a web page, attach to my Chrome, CDP, remote debugging port, connect_over_cdp, browser automation, or scrape a page I am logged into. NOT for a one-off look at a tab the owner already has open (use claude-in-chrome) and NOT for fetching a public page (use WebFetch or firecrawl).
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
blueprint
Use when blueprint is needed. Turn a one-line objective into a step-by-step construction plan for multi-session, multi-agent engineering projects. Each step has a self-contained context brief so a fresh agent can execute it cold. Includes adversarial review gate, dependency graph, parallel step.
biomedical-paper-triage
Use when turning a raw list of biomedical papers into a ranked, access-tagged shortlist graded T1-T4 by human/animal/in-vitro evidence strength via Scholarly Access MCP ; do NOT use for generic literature metadata or citation graphs (use scholarly-literature-review) ; returns defensible ranked shortlist with evidence grades.
binja-sigmaker
Use when generating byte signatures or AOB patterns from Binary Ninja for runtime hooks, trainer work, or version-resilient function identification.