Claude Skills
The open format is called Agent Skills and works in Claude Code, Codex, Cursor and other agents — most people know it as Claude Skills.
Every Agent Skill we could find on GitHub, deduplicated by content. 79 870 files from 1 769 authors, of which 62 217 are unique — the rest is the same skill repackaged into someone else's repository. For each one: what it weighs in tokens, whether it ships runnable scripts, and which MCP servers it needs.
Answer-first writing using the Minto Pyramid Principle. Lead with the conclusion, then 2–4 MECE supporting arguments, then evidence under each. Use whenever you have a recommendation, finding, or assessment to communicate in writing — memos, emails to execs, Slack updates that need to land, briefing docs. Different from the Decision Memo Builder (which is a specific 1-page decision artifact); this is the general writing method.
Design a half-day or full-day strategy workshop that ends with a committed decision and named owners — not a "great discussion" with no artifact. Outputs an hour-by-hour agenda, breakout prompts with expected outputs, pre-reads, facilitator playbook for when things go sideways, and the post-workshop artifact that justifies the day. Use for offsites, leadership alignment sessions, kickoff workshops, planning meetings >2 hours.
Map the people who can make or break a decision onto a Power/Interest grid, with each named stakeholder's stance (champion / skeptic / blocker / unknown), what they care about, and the single influence move you'll make for them. Use before any cross-functional decision, big launch, reorg, or executive ask — anywhere the politics will decide the outcome more than the merits.
>- 生成小红书(RedNote)笔记：把主题或用户提供的素材(文字/图片/视频)做成一套竖版卡片(图文) 或一段带分镜脚本的视频稿(视频)，配 caption + hashtags，落盘到 output/小红书/。 使用时机：用户说 写小红书 / 做小红书笔记 / 小红书图文 / 小红书种草 / 出一套 xhs 卡片 / 小红书视频 / write a Xiaohongshu note / make a RedNote post。
Keep dependencies up-to-date. Discovers outdated deps via dependabot alerts/PRs, creates one PR per ecosystem, iterates until CI is green, then assigns for review.
Run strict OpenShift API review workflow for PR changes or local changes
Generates a standardized release note description for an issue, commit, or pull request. Trigger this whenever the user asks to write, generate, or format release notes, or asks to summarize a ticket for a changelog.
Manage local Markdown-first My Wiki knowledge vaults with an AI agent. Use for capturing webpages, PDFs, Office documents, notes, images, folders, and ZIP bundles; maintaining raw-to-wiki evidence links; searching or answering from a vault; checking or repairing vault health; switching among local vaults; and opening the My Wiki knowledge graph or dashboard.
China antique parchment MapLibre map engine with configurable basemap, Terrarium hillshade, optional user-supplied water overlay, live tuner page, and tiered settlement extrusions for HyperFrames historical videos.
Develop and evaluate logo concepts when the explicit design problem is semantic fusion: two or more brand meanings must share a contour, stroke, negative space, glyph skeleton, or shape system. Use for fusion-focused design or redesign, critique, comparison, or prompt-only work on a logo, symbol, app icon, monogram, or wordmark. Do not use for generic logo creation, general identity systems, illustration, packaging, or type selection unless semantic fusion is a stated or clearly central requirement.
Generate new raster assets that must contain native pixel transparency, then verify the untouched PNG or WebP before delivery. Use for transparent stickers, sprites, character cutouts, product assets, and furry or fine-edged subjects where background removal would be unacceptable. Do not use to remove a background from an existing opaque image, repair an opaque generation by masking or chroma key, or create ordinary opaque images.
Create, audit, and repair short page-native diary comics around an existing authorized recurring character, with story-directed page rhythm, exact dialogue, directional-surface proof, and original-resolution visual QA. Use when extending an established recurring-character series from an anecdote, conversation, dream, or viewpoint, or when checking or repairing identity, anatomy, text, prop, relation, continuity, or layout defects. Do not use for standalone character design, educational explainers, infographics, one-off illustrations, generic memes, unauthorized characters or marks, imitation of a named living artist, watermarking, or social-platform publishing.
Expo React Native mobile app development with RevenueCat payments, AdMob ads, i18n localization, onboarding flow, paywall, and NativeTabs navigation
Use when converting academic paper PDFs or structured text into professional, high-standard PowerPoint (PPTX) presentations for defense or seminars.
最好的找Skill的方式，能够基于你的任务，去寻找最匹配的高质量Skill。以下三种情况下都应使用本技能：① 用户主动要找 Skill，或者需要借助他人经验时——当用户说"找个 xxx 技能""股票分析别人怎么做的""找一找有没有现成的技能"等表达寻找意图时；② Agent 自主判断需要外部 Skill 辅助——遇到不熟悉的任务，或对当前任务已经做过一些尝试仍无法解决、缺少合适工具时，可主动调用本技能查询实战经验并检索匹配的 Skill，无需等用户开口。；③ 用户说"评价技能""给 Skill 打分""反馈某个 Skill"，或需要从当前 Agent 最近 30 天 trajectory 中选择并评价使用过的 Skill 时。评价支持已知 trajectory Provider，也支持当前 Agent 自定位和动态理解未知格式。
> Use when composing the Search(...) call and calling .start(). Covers the four worker classes (MultiProcessWorker / IsoMultiProcessWorker / RayWorker / AsyncWorker) and when to pick each, SearchConfiguration sizing rules, dump_results checkpointing, tracker_config choice (Disabled / Loguru / MLflow), num_workers/task_timeout semantics, and GPU clock locking for stable measurements. Triggers on "Search()", "tuner.start()", "pool_size", "num_workers", "task_timeout", "IsoMultiProcessWorker", "RayWorker", "dump_results", "MLflow", "GPU clocks".
> Use AFTER a Search has completed and BEFORE claiming any speedup or shipping an ACF. Loads the dump_results CSV, extracts top-K candidates (single-objective) or the Pareto front (multi-objective), re-measures each against the no-ACF baseline with 100+ trials on fresh caches, runs Welch's t-test plus Cohen's d, rejects three classic false-positive patterns (lucky-min / higher-variance / multiple-comparisons-of-N), and saves the validated winner as best.acf. Triggers on "validate result", "extract best config", "Welch's t-test", "is my speedup real", "save best ACF", "pareto front", "claim speedup", "ship config".
> Use when starting a fresh CompileIQ project, hitting a socket timeout, or before running any other compileiq-* skill. Verifies CUDA 13.3+, ptxas, GPU access, that `from compileiq.ciq import Search` and friends resolve, and that `PtxasSearchSpace().retrieve()` returns a real path. Documents the env vars that control timeouts, caching, and search-space mirroring. Triggers on "set up compileiq", "compileiq doesn't work", "socket timeout", "where do search spaces come from", "air-gapped compileiq".
> INVALID_SCORE, scores aren't improving, every config returns the same number, ptxas errors fill the log, CV% is too high, or a winning ACF candidate needs NCU profiling to explain. Symptom-indexed table on top. Triggers on "compileiq hang", "socket timeout", "INVALID_SCORE", "not converging", "every score is the same", "TypeError fromhex", "ncu profile", "register spill", "ptxas error", "not in expected format", "high cv".
> Use BEFORE running a full CompileIQ search. Walks through downloading a Booster Pack from NVIDIA/CompileIQ GitHub Releases, applying ACF candidates one at a time to the user's compiler (raw PTXAS, NVCC, Triton, Helion, FlashInfer), and keeping only candidates that compile, pass correctness, and beat the no-ACF baseline. Includes the mandatory Debug-pack O0/O3 ACF-injection canary that proves the ACF is reaching PTXAS. Triggers on "booster pack", "ACF", "apply-controls", "speed up without searching", "helion fp8", "flashinfer batch decode", "debug pack".
> Use when picking the search_space= argument for Search(). Covers the three provider classes (PtxasSearchSpace, NvccSearchSpace, LocalSearchSpaceBin), how to pin a version/variant/tag, the attention-focused 'att' variant for attention kernels (FlashAttention, GQA, MHA, MLA, FlashInfer Batch Decode), air-gapped mirroring via CIQ_SEARCH_SPACES_DIR, and custom user-defined search spaces built from compileiq.search_spaces.base primitives. Triggers on "search space", "PtxasSearchSpace", "NvccSearchSpace", "air-gapped compileiq", "offline compileiq", "CIQ_SEARCH_SPACES_DIR", "attention variant", "ptxas att".
> Use when writing the objective_function= passed to Search(). Covers the two legal signatures (compiler-only str vs mixed list), the baseline-knockout branch, per-eval cache busting, framework-specific --apply-controls injection (raw PTXAS, NVCC, Triton, Helion, cuTeDSL/FA4, FlashInfer), correctness-before-timing, INVALID_SCORE handling, and the Debug-pack O0/O3 ACF-injection canary that must pass before launching a search. Triggers on "objective function", "apply-controls", "INVALID_SCORE", "save_compiler_config", "baseline knockout", "BASELINE_CONFIG", "every config returns the same score", "TypeError fromhex".
>- ALWAYS invoke this skill before asking the user any technical question or offering options, and whenever they ask to be asked in plain words - "ask simple", "ask me simply", "ask me in plain words" - or ask if something is worth doing or if something simpler would do, in any language. Rewrites the question in plain words and always ends with one marked recommendation. A simpler substitute, extra work later, lock-in, over-engineering, easy to undo. Doing nothing is always weighed.
>- ALWAYS invoke this skill when work wraps up or the user asks how it went - "done or not", "are we done", "what happened", "report" - in any language, and when a Stop hook asks for a session report. Produces a ten-line plain- anything needed from you, new debt, safe to close. Every "yes" names its proof; unverified says "not checked". Saves the report so the next session starts from it instead of re-exploring the repo.
>- ALWAYS invoke this skill when the user asks for simpler or shorter about something said or written - "say it simply", "what does this mean", "I don't understand your answer", "too long", "wait, what?", "bro" - in any error. "I don't understand what to DO" is os-step-by-step; this skill the point, keeps every number, warning and caveat - no facts added, no bad news dropped. A number returns exactly that many points, most important first.
>- ALWAYS invoke this skill when the user asks about work done outside this session - "check work", "check the others", "check other sessions" - or to each part against machine state, name every gap between claimed and true. Verified-ready work merges in the same pass. End with what is left, one next step, and any reply another session needs, ready to paste.
>- ALWAYS invoke this skill before anything hard to undo gets agreed to - a contract, a purchase, a migration, a launch, a price change, a reorganisation - and whenever the user asks "what could go wrong", "what are we missing", "poke holes in this", or for a premortem or a red team, in any language. Assumes the decision already failed and works backwards to find out why, in a fresh agent that had no hand in making it. Sweeps nine areas go ahead, go but fix these first, try it small first, think again, do not do this.
>- ALWAYS invoke this skill when you need the user to act - run a command, paste a secret, click, approve - and whenever they ask how to do something "what do I do", "what should I do", "I don't understand what to do", "explain what I need to do", in any language. Picking which task comes next is os-whats-next; this skill is for doing the thing in front of you. First only they can do. Then one action per step, commands labelled by what they touch, no jargon. Commands are single lines that prompt for any value - typing hidden for secrets - and confirm in plain words. Afterwards verify their step.
>- ALWAYS invoke this skill when the user asks what to do next, what is left, or what is blocked - "what's next", "what now", "what should we work on", "anything I can do" - in any language. This skill picks the next piece of work; when the user asks HOW to do a thing or says they do not understand what to do, that is os-step-by-step. Reads the last report, local changes, verified-ready pull requests merge in the same pass. Then sorts the rest into doable-alone and needs-you, ending with one recommended next task in plain words - what it closes or unblocks. Never invents tasks.
Generate grounded NPC intentions, dialogue, memory and optional movement policies for Infinite Pokémon from a supplied saved-run observation or map-design context.
Set up and start Infinite Pokémon when a user asks to install, play, or launch the game with their AI assistant. Clone missing game files, install dependencies, build, and launch the local desktop or browser game; not for background map or NPC generation jobs.
Generate Infinite Pokémon map stories, terrain features, content profiles, NPC policies, and interiors from a host-supplied saved-run snapshot and output schema.
Design interior furniture layouts, rugs, room names and inspectable details for new Infinite Pokémon maps within the supplied protected room and service constraints.
Find new skills on GitHub or check a specific skill before installing. Use when the user wants to search for skills, evaluate a skill URL, check overlap and security risk before installing, or compare a local skill against alternatives. Trigger with '/janitor-discover'.
Heuristic security scan of installed skills — prompt-injection phrases, hidden unicode instructions, credential-store access, network-pipe-to-shell and payload-smuggling patterns. Use when the user asks 'are my skills safe', wants to scan skills for prompt injection or malware patterns, or before trusting a newly installed skill. Trigger with '/janitor-security'.
Tinder for your Claude Code skills. Reviews a sorted deck of every installed skill and lets you swipe keep / delete / skip on each one. Use when the user wants to bulk-clean their skill collection, triage unused skills, or do interactive skill cleanup. Trigger with '/janitor-swipe'.
Full health check of all your skills in one report. Use when the user wants to check for errors, find duplicates, detect broken skills, or get a complete overview of skill health. Pass --brief for inventory only. Trigger with '/janitor-report'.
Automatically fix skill problems (safe preview first). Also use with --prune to find and remove broken symlinks, empty directories, and orphaned skills. Trigger with '/janitor-fix'.
Show whether each skill is earning its context-window cost — combined tokens-used view sorted by waste. Use when the user asks 'are my skills worth it', 'what's my context budget', 'which skills are dead weight', or wants to audit skill value, token cost, or usage. Trigger with '/janitor-value'.
Agent-callable Algolia tools — index records, search and browse them, manage index settings, synonyms, and query rules, and read AI recommendations.
Agent-callable ElevenLabs tools — generate spoken audio from text, create sound effects and multi-speaker dialogue, re-voice and clean up audio, transcribe audio and video, design synthetic voices, and manage voices, history, and quota. Use when the user mentions ElevenLabs or wants AI audio work — text to speech, narration, voiceover, transcription, voice changing, or sound design — even if they don't name ElevenLabs explicitly, e.g. "read this aloud", "make an MP3 of this", "transcribe this recording".
Agent-callable Dropbox tools — upload, organize, find, and share files and folders. Use when the user wants to manage Dropbox content (save, move, search, share, list, or read files), including requests that don't name Dropbox explicitly, e.g. "save this report to my cloud storage" or "share that folder with Sam".
Agent-callable Google Ads tools — search campaigns, ad groups, and ads via GAQL, build performance reports, manage campaign status and budgets, and set up conversion tracking. Use when the user mentions Google Ads or wants to read or manage advertising campaigns, budgets, or reporting — even if they don't name Google Ads explicitly.
Agent-callable Discord tools — send and manage messages, start threads and forum posts, manage channels, members, and roles, and post via webhooks. Use when the user mentions Discord or wants to post, read, or manage Discord servers, channels, or members, even if they don't name Discord explicitly.
Agent-callable Alpaca trading tools — place and manage stock, crypto, and options orders, read account balances, positions, and portfolio history, look up assets and market hours, and read watchlists. Use when the user mentions Alpaca or wants to trade or inspect a brokerage account, even if they don't name Alpaca explicitly.
Agent-callable DataForSEO tools — Google SERP results, keyword and domain analytics, backlinks, Google Maps business data, on-page audits, and AI-search visibility (LLM answers + brand mentions). Use when the user wants SEO or AI-search data, even if they don't name DataForSEO.
Agent-callable Clay tools — create, update, and find rows in Clay tables, and navigate workspaces, tables, views, and users. Use when the user mentions Clay or wants to add, update, or look up rows in a Clay table, even if they don't name Clay explicitly.
Agent-callable Google Analytics 4 (GA4) tools — run analytics reports, discover the dimensions and metrics a property supports, navigate accounts and properties, manage key events and custom dimensions/metrics, and send Measurement Protocol events. Use when the user mentions Google Analytics or GA4, or wants website/app traffic, conversion, or reporting data — including requests that don't name GA4 explicitly, e.g. how many users visited last week, top pages by country.
Agent-callable Google Contacts tools — create, find, update, and delete contacts, manage contact groups (labels) and membership, and read auto-saved other contacts. Use when the user mentions Google Contacts or wants to look up, save, or organize people — including requests that don't name Google Contacts explicitly, e.g. add Jane to my contacts, find Bob's email.
Agent-callable Google Calendar tools — create, update, move, search, and delete events, manage calendars, check free/busy availability, resolve event colors, and manage calendar sharing. Use when the user mentions Google Calendar or wants to schedule, find, reschedule, or share events and calendars — including requests that do not name Google Calendar explicitly, e.g. "put a meeting on my calendar Tuesday 3pm" or "am I free Friday afternoon".
Agent-callable Google Sheets tools — read and write spreadsheet data as rows or raw cells, manage worksheets and columns, and apply formatting, sorting, and validation. Use when the user mentions Google Sheets or wants to read, add, update, look up, or organize spreadsheet data, even if they don't name Sheets explicitly.
Agent-callable Google Tasks tools — create, list, update, complete, move, and delete tasks and task lists. Use when the user wants to manage Google Tasks or to-dos, even if they don't name Google Tasks explicitly.
Agent-callable HeyGen tools — generate AI avatar videos, translate and lip-sync videos, synthesize speech, and clone or browse voices and avatars. Use when the user mentions HeyGen or wants to create, translate, or check the status of AI videos, or to generate or clone a voice — including requests that don't name HeyGen, e.g. make an avatar video from a script, translate a video to Spanish, turn a script into a voiceover, or clone a voice from a recording.
Agent-callable Google Docs tools — create, read, export, and find documents; edit text, formatting, images, and page style. Use when the user mentions Google Docs or wants to create, read, search, or edit document content, even if they don't name Google Docs explicitly.
Agent-callable Microsoft Outlook tools — read and search mail, send/reply/forward, organize messages and folders, manage calendar events, and manage contacts. Use when the user mentions Outlook, Microsoft 365 mail/calendar/contacts, or wants to send, read, search, or organize email, schedule events, or look up contacts — even if they don't name Outlook explicitly.
Agent-callable Microsoft SharePoint tools — find sites and document libraries, browse, search, upload, move and share files, manage lists and list items, and author site pages. Use when the user mentions SharePoint or wants to work with SharePoint sites, files, folders, lists, or pages, even if they don't name SharePoint explicitly.
Agent-callable Notion tools for searching pages and databases, reading and creating pages, querying data sources, appending content, and managing schemas. Use when the user mentions Notion or wants to find, read, create, or edit Notion content, even if they don't name Notion explicitly.
Agent-callable Resend tools — send transactional email, manage contacts and segments, send broadcasts, read delivery status, and diagnose sending domains. Use when the user mentions Resend or wants to send email, manage a contact list, or check email delivery, even if they don't name Resend explicitly.
Agent-callable Trello tools — create and update cards, manage boards, lists, labels, checklists, and search. Use when the user mentions Trello or wants to create cards, move tasks, or manage boards, even if they do not name Trello explicitly.
Create, find, complete, and organize tasks and task lists in Microsoft To Do (Microsoft's to-do app), plus break tasks into checklist steps. Use when the user wants to capture a task, check their to-do list, mark something done, or manage their tasks/lists — even if they don't say "Microsoft To Do" explicitly (e.g. "add buy milk to my tasks", "what's on my to-do list", "mark that done").
Claude Skills — questions
Answers built from the skills we actually parsed.