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.
Run local Go health checks (tests, coverage, lint) in Go repositories that contain go.mod/go.sum. Use when the user asks to run or interpret local Go test/coverage/lint workflows using tools like lazygotest, gocovsh, tparse, and golangci-lint. Do not use for Rust or non-Go projects.
Run a reusable JavaScript supply-chain security baseline with pnpm-first hardening, release-age gating, lifecycle-script controls, exotic dependency checks, CI install checks, and optional incident IOC profiles.
Concise git workspace snapshot for the current directory. Use when asked to show current branch, cwd, repo root, whether the current directory is a worktree, local dirty status, or whether other worktrees have uncommitted changes.
Use the no-mistakes gated push workflow in Git repositories. Use when the user wants to check no-mistakes readiness, initialize or inspect a gate, configure .no-mistakes.yaml, push through the no-mistakes remote, attach to runs, interpret/fix pipeline findings, rerun status/runs/doctor, or use no-mistakes with Codex/Claude/OpenCode/Pi/ACP agents. Install guidance is secondary and only for missing or broken no-mistakes binaries.
Publish CLIs/TUIs to Homebrew via a personal tap. Use when asked to create or manage a Homebrew tap repo, generate or update formulae, compute sha256, test installs, or ship new releases for Go, Rust, Node/TypeScript, Python, or prebuilt binaries.
Performs root-cause-first debugging and review by tracing expected behavior to the first unintended side effect before changing contracts, parsing, or types. Use when debugging protocol errors, deserialization failures, null payloads, missing fields, restore or hydration issues, state-ownership bugs, unexpected requests, background mutations, or reviewing junior-created code where the visible failure may be downstream noise.
Run or install repo security leak checks with BetterLeaks and Trivy. Use when asked to scan for leaked secrets, vulnerable dependencies, misconfigurations, add secret-leak guardrails, add BetterLeaks, add forbidden-path hooks, or run secleak-check before release.
Stage a finished local feature, run local verification, create a Conventional Commit, then send it through the no-mistakes gated review/fix loop before real upstream push/PR. Use when the user says a feature is done, asks for a professional stage-review flow, wants Codex to git add and commit safely, or wants no-mistakes review/verify/fix before pushing to origin.
Find, clone, inspect, and summarize high-quality GitHub reference repositories for coding agents. Use when a user asks for GitHub reference projects, examples, prior art, inspiration, implementation patterns, or includes "$search-context" in a coding prompt.
Run shellcheck on shell scripts after editing scripts or when debugging shell errors. Use for linting scripts in a repo (especially scripts/), catching issues like set -u with unset vars, bad subshell usage, or quoting mistakes.
Start the local webpack dev server and connect to it with the Playwright browser. Use this skill when asked to start the dev server, preview the app, or interact with the running application in a browser.
Build the Cosmos Explorer project. Use this skill when asked to build, compile, or check the build status of the project.
Run unit tests for the Cosmos Explorer project. Use this skill when asked to run tests, check test results, or debug test failures.
Map this codebase into project memory — a code graph of every symbol and how they connect, plus a written profile of stack, layout, conventions and workflows. Re-run any time to refresh both in place. Use when memory is newly installed on an existing project, or when the project has changed enough that the stored map is stale.
Restructure agent instruction artifacts (Claude Agent Skills, rules files, CLAUDE.md, playbooks) by applying five structural principles derived from the Open Knowledge Format - bounding the worst-case context a question loads (the mean may rise; the win is the tail and answer accuracy) and making cross-references survive rewrites. Use this skill whenever the user wants to split a monolithic skill or reference file, reduce context loaded per question, add an index layer, make cross-references survive rewrites, or audit a skill's structure for efficiency. Also use it when the user mentions progressive disclosure, token-efficient skills, or restructuring instructions for agents.
| Deploy a working VLESS + XHTTP + TLS + Cloudflare CDN VPN service on a fresh VPS installing x-ui/3x-ui/x-ui-pro, configuring VLESS/XHTTP/Reality, self-hosted VPN, bypass GFW, anti-censorship, 科学上网、翻墙、搭建代理、自建梯子、VPN 节点部署、 Cloudflare CDN 代理配置。Use this even if the user originally asks about other Trojan, Trojan-Go, V2Ray WebSocket, Hysteria, WireGuard, OpenVPN, Outline, Algo VPN, sing-box, NaiveProxy — when the goal is bypassing the Great Firewall VPN connection failures, adding clients/users, certificate renewal, Xray config debugging, panel access via SSH tunnel, firewall adjustments, VPN 连不上排查、 添加客户端、证书续期、面板访问。Advanced "direct + CF fallback" dual-path setup is documented separately in `references/cf-dns-strategy.md` as an opt-in optimization after the basic deploy works.
>- Shoot, animate, title and optionally score product demo videos with Remotion — for a landing page, docs, onboarding, an app store listing or a social post. Use when editing anything under src/compositions/, when the user asks for a new demo clip, an intro or an animated title, asks for sound or a version for social, says a clip feels dead or looks like a screen recording, or reports tiling, flashing or flickering in a rendered video.
v2.4.0 · Dungeon Master assistant for running persistent D&D 5e campaigns. Handles campaign creation/loading, character management, combat tracking, NPC generation, dice rolling, and session state — all persisted across sessions. Invoke with /dm:dnd followed by a subcommand, or just speak naturally once a campaign is loaded.
Delegate tasks to Claude Code CLI for prototyping, debugging, and code review. Supports multi-turn sessions via SESSION_ID.
Turn a large task into a persistent plan file and a trackable Issue CSV (local files, not an external issue tracker), then execute the issues autonomously with per-row status updates. Use when work must survive across sessions or hand off between agents, when you need an auditable record of multi-step execution, or to resume a plans/ and issues/ pair from an earlier session. Do not use for single-session tasks that your harness's built-in plan or todo tracking already covers.
Delegate tasks to Codex CLI for prototyping, debugging, code review, implementation handoff, cross-model second opinions, and multi-turn Codex sessions via SESSION_ID.
Delegate tasks to Google's Antigravity CLI (agy) for prototyping, debugging, code review, and research. Supports multi-turn sessions via SESSION_ID. Successor to the retired Gemini CLI.
Delegate tasks to Grok CLI (xAI) for prototyping, debugging, code review, research, and cross-model second opinions. Supports multi-turn sessions via SESSION_ID.
> Low-noise self-improvement skill. Captures learnings, errors, and corrections into a local `.learnings/` cache layer. Never auto-promotes to long-term memory files (SOUL.md, MEMORY.md, AGENTS.md, TOOLS.md). Promotion requires explicit user approval after repeated validation.
Use when drafting, rewriting, reviewing, or planning Chinese long-form content for public accounts, blogs, newsletters, professional analysis, product or technical writing, especially when preserving an author's voice, using personal material, or reducing formulaic AI-sounding prose.
HTML/PPT 风格网页演示文稿生成工作流。用户提供第一页 Hero Prompt、现有 Hero 工程、演示文稿扩展指令、或要求"先对齐再执行"、"需要你更改的部分: {}"、"把 hero 扩成多页汇报/slide deck/HTML PPT"时必须使用。本 skill 要求先同步理解、列出需要用户更改或确认的部分，得到明确确认后才修改代码，并保持第一页视觉系统一致。
把需求、原始 PRD、竞品分析、零散笔记或模糊产品想法提炼成可直接发给 Claude Code、Codex 或其他编码 Agent 的精简中文 Markdown PRD。触发：用户说“帮我写一个用来vibe coding的PRD”、要求整理 coding-agent-ready PRD/spec/prompt，或希望从业务/产品材料中抽取实现所需信息。
Clawd 像素风动画生成器。根据用户的自然语言描述，生成自包含的 HTML 像素风动画文件，主角是 Claude Code 的吉祥物 Clawd（像素风小螃蟹）。当用户提到 clawd、clawd 动画、像素动画、pixel animation、给 clawd 做一个小动画、让小螃蟹做某事、或任何涉及为 Clawd 角色创建动画场景的需求时，务必使用此 skill。即使用户只是说"让小螃蟹安慰一下我"或"今天和小螃蟹的合作很愉快"，也应触发。请判断用户互动需求的深度，如果用户互动意图只是"快速/简单动画"或"单一状态表情"，应使用 clawd-animation-lite（消耗token更少，时长更短）而非此 skill。
协助意疏写「意疏的AI口袋」公众号文章。Claude 是协作者不是写手——voice 必须从意疏嘴里出来，AI 负责结构化、扩写、降 AI 味。触发：写公众号文章 / 帮我整理口述 / 有素材但想不清 / 这段 AI 味太重 / 看看有没有意疏味道 / 帮我把骨架写成文章。输出 Markdown，排版找 wechat-formatter，发布找 wechat-publisher。
Clawd 像素风动画生成器（轻量版）。根据用户的自然语言描述，生成1-3秒的简短clawd动画，当用户的需求的互动场景较为具体和单一，请采用轻量版本生成token消耗少、生成速度快的HTML像素风动画文件。主角是 Claude Code 的吉祥物 Clawd（像素风小螃蟹）。当用户提到 clawd、clawd 动画、像素动画、pixel animation、给 clawd 做一个小动画、让小螃蟹做某事、或任何涉及轻量化的为 Clawd 角色创建动画场景的需求时，务必使用此 skill。即使用户只是说"让 clawd 开心一下"或"做个小螃蟹的动画"，也应触发。
HTML/React 第一页 Hero 生成工作流。用户提供 Hero Prompt、品牌主题、首页视觉需求、背景视频/图片、或要求"先对齐再生成第一页"、"需要你更改的部分: {}"、"做一个能跑的交互 Hero 页面"时必须使用。本 skill 要求先同步理解和需要替换的部分，确认后再创建或修改前端工程，并交付可运行、可预览、可打包的第一页。
把 Markdown 文稿转换成微信公众号可用的内联样式 HTML。**触发场景**:用户说 "把 markdown 转成公众号能用的 HTML" / "排版一下这篇文章" / "套个主题" / "微信编辑器不识别我的样式" / 任何提到「公众号排版 / 微信 HTML / md 转 html / doocs 主题」的请求。微信编辑器不识别 `<style>` 标签,必须把所有 CSS 内联到每个 HTML 元素的 style 属性里——本 skill 用 doocs/md 主题完成这件事。本 skill 只负责格式转换;写作交给 wechat-writer,发布交给 wechat-publisher。
微信公众号文章的发布管道——封面生成、本地图片上传素材库、创建草稿、发布草稿、以及从 mp.weixin URL 提取原文。**触发场景**:用户说 "发到公众号草稿箱" / "推送到公众号" / "做张公众号封面" / "上传图片到微信" / "把这篇文章提取成 markdown" / "改写这篇 mp.weixin 链接(的提取部分)" / 任何提到「发布 / 草稿 / 封面 / 推送 / 提取微信文章」的请求。本 skill **不写文章**(交给 wechat-writer),**不做 markdown → html 排版**(交给 wechat-formatter)——它只处理图片生成、文件上传、微信 API 调用。
Manage Claude Code session transcripts from local JSONL storage, including listing candidate sessions, exporting full or selected sessions to organized Markdown, inspecting archives, and summarizing tool-call history. Use when the user asks to scan, parse, archive, inspect, recover, summarize, manage, or convert Claude Code sessions, `~/.claude/projects` data, `.jsonl` transcripts, tool-call history, or hard-to-read Claude Code conversation logs.
Generate elegant cover images for articles. Analyzes content and creates eye-catching hand-drawn style cover images with multiple style and composition options. Use when user asks to "generate cover image", "create article cover", or "make a cover for article".
Add Feishu (飞书/Lark) as a channel. Uses WebSocket long connection mode — no public URL needed. Bot responds to all messages (no trigger required).
Convert Mermaid diagrams and Markdown tables to images (PNG) for platforms that don't support rich formatting. Use when user asks to "convert to image", "export as PNG", "make this an image", or has content for X/Twitter that needs visual exports.
Fantasy Premier League copilot: syncs live FPL data, analyzes players/teams/fixtures, manages your fantasy squad, and generates self-contained HTML reports for squads, fixtures, transfers, captain picks, and gameweek strategy. Use when the user asks about FPL, player stats, transfer advice, captain picks, fixture difficulty, gameweek strategy, or squad management.
Use when an agent needs to delegate a task to the OpenAI Codex CLI from another agent environment such as Claude Code, OpenClaw, or similar. Covers checking whether Codex CLI is installed, running one-off Codex prompts with `codex exec`, resuming sessions, collecting outputs, attaching images or files as input with `-i`/stdin, and handling Codex image generation including finding and reporting generated image file paths.
Design typography-driven video cover images using HTML/CSS + Chrome DevTools screenshot. Generates covers in all needed aspect ratios - 16:9 (YouTube), 16:10 (Bilibili), 9:16 and 3:4 (抖音/视频号 竖屏短视频) - with big readable text. Different from `cover-image` (AI hand-drawn aesthetic) - this is precise typography control via code. Use when user asks for "视频封面", "thumbnail", "做封面", "cover design", "缩略图", "横屏/竖屏封面", "抖音封面", "视频号封面".
Claude Skills — questions
Answers built from the skills we actually parsed.