AI Skills Hub - Browse 180,000+ Agent Skills

マルチブランチ並列開発オーケストレーター。複数のClaude Codeインスタンスを部下として管理し、並列でタスクを実行させる。自然言語で指示を出し、進捗管理・評価・マージを行う。

Create well-structured git commits following conventional commits format. Use when the user asks to commit changes or when finishing a piece of work.

Create or update project-specific skills in this repo. Use when asked to make new skills, modularize workflows, or build skill packs for this codebase.

Stage, commit, and push changes to GitHub

扫描 GitHub 仓库或本地目录，发现符合 Anthropic 标准的 Skills，与本地 Skills 库进行对比，并支持一键安装新技能。适用于发现和引入新的 Agent 能力。

Search technical documentation using executable scripts to detect query type, fetch from llms.txt sources (context7.com), and analyze results. Use when user needs: (1) Topic-specific documentation (features/components/concepts), (2) Library/framework documentation, (3) GitHub repository analysis, (4) Documentation discovery with automated agent distribution strategy

Use when user requests Gemini to analyze multimodal content (images, PDFs, audio, video), process long documents, or scan large codebases - dispatches to gemini-executor agent for isolated execution

server 모듈에 TDD(Test-Driven Development) 방식으로 기능을 구현합니다. 새로운 기능 개발, 버그 수정, 리팩토링 시 사용합니다.

Summarize CI test reports with intelligent signal detection. Distinguishes flaky tests from real failures, surfaces only critical/stale bugs, and provides actionable deployment recommendations.

Verify legal basis using e-Gov Law API. Search Japanese laws (income tax, labor, corporate) for articles on withholding tax, social insurance, employment regulations. Use when user mentions laws, legal requirements, tax regulations, or requests accurate legal references.

Design scalable Express.js REST APIs. Focuses on controller/service/repository layering, consistent response formatting, error handling, and API versioning. Use when creating new endpoints, refactoring monoliths, or establishing backend standards.

Implement robust technical SEO for Next.js applications. Covers Metadata API, Canonical URLs, Custom Sitemap/Robots generation scripts (user preference), and GA4/GSC integration. Focuses on crawlability, indexing control, and analytics accuracy.

Convert API documentation to Protocol Buffer (.proto) definitions. Use when given a URL to API docs (OpenAPI, Swagger, HTML docs, JSON examples) and asked to generate protobuf message types for request/response payloads. Triggers include requests to create .proto files from APIs, convert API schemas to protobuf, or generate typed message definitions from documentation.

狀態保存與恢復知識。自動載入於需要保存進度、恢復狀態、長時間任務相關情境。 觸發詞：checkpoint, 檢查點, 狀態, state, 保存, save, 恢復, restore, resume, 進度, progress

錯誤處理與恢復策略知識。自動載入於遇到錯誤、異常、失敗需要處理的情境。 觸發詞：error, 錯誤, exception, 異常, fail, 失敗, retry, 重試, fallback, 回退, recover, 恢復

Discovers and executes MCP tools on-demand via agentsbox_* tools. Use when you need to find the right tool, when a user asks to use a tool you don’t have in context, or when troubleshooting MCP/tool availability.

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

Search and fetch photos from Unsplash with proper attribution. Use when users need images for content, want photos by keyword, need random images for variety, or ask for stock photos. Automatically provides attribution required by Unsplash guidelines.

Create, read, write, format, and manage Google Sheets spreadsheets

Control browser via BrowserOS MCP - use JavaScript and screenshots together for reliable extraction

Send, read, search emails, manage labels, drafts, and attachments

List, search, download, upload, and manage Google Drive files

PDF generation, extraction, and document conversion tools

Refactor legacy or messy code without breaking functionality. Focuses on creating a safety net (tests) first, then applying small, verifiable structural changes.

ToC auto-generation toolkit for rules_toc.yaml and specs_toc.yaml. Provides scripts for pending YAML generation, merge processing, validation, and checksum management.

Control Claude Code notification system with sound and window highlighting

新規記事ファイルを作成するスキル（Zenn/Qiita両対応）。 「新しい記事を作成」「記事を書く」「Zenn記事を作成」「Qiita記事を作成」などのリクエストで使用。

LinkedIn positioning and content strategy for product design studios. Use when building LinkedIn presence, creating content, attracting clients, or developing thought leadership for industrial design firms and creative consultancies.

Install Codex skills into $CODEX_HOME/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).

What it does and when to use it. Max 1024 chars.

EEW（緊急地震速報）機能のアーキテクチャ解説。EEW関連のコード修正、予報電文・警報電文の処理フロー、キャッシュ管理、ソース優先順位の理解に使用。

Decompose large unified specifications into agent skills with progressive disclosure. Use when converting documentation, frameworks, or knowledge bases exceeding 50KB into properly structured skills that Claude can navigate efficiently.

Build algorithmic trading strategies with QuantConnect LEAN engine. Use when creating trading algorithms, backtesting strategies, implementing technical indicators, setting up live trading, or using the LEAN CLI. Covers Python algorithm development with support for equities, options, futures, forex, and crypto.

Extracts Issue ID from the current branch name. Use this skill when you need to determine the Issue number associated with the current work context.

Convert PDF files to Markdown format with optional image extraction. Use when you need to extract text from PDFs, convert PDFs to Markdown, or extract images from PDF documents.

Use when tests have race conditions, timing dependencies, or inconsistent pass/fail behavior - replaces arbitrary timeouts with condition polling to wait for actual state changes, eliminating flaky tests from timing guesses

Use when resolving Steam library paths, constructing compatdata paths, or working with secondary Steam libraries. Covers path construction, validation order, and common pitfalls.

Use after completing code changes. Covers pytest commands, mypy type checking, black/isort formatting, and the complete verification checklist before marking tasks complete.

Help design and add scenarios (Given-When-Then specifications) to slices. Use when a user wants to specify behavior for a slice, add test cases, or define invariants.

Create a complete state-change slice with Screen, Command, Event, flows, and field mappings. Use when a user action triggers a state change in the system.

PR/コミット/ステージ差分のDIFF取得手順

devbooks-docs-consistency：检查并维护项目文档与代码的一致性，支持增量扫描、自定义规则与完备性检查。可在变更包内按需运行或全局检查。旧名称 devbooks-docs-sync 保留为别名并输出弃用提示。

Convex module structure conventions, validation, and engineering decisions

Lifecycle manager for GitHub-based skills. Use this to batch scan your skills directory, check for updates on GitHub, and perform guided upgrades of your skill wrappers.

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

三层定位抓取 Skill（正则→Embedding→LLM），用于精准定位文档章节

Use when working with Daytona SDK - uploadFile(Buffer, path), Bun install, CodeLanguage options

Production monitoring навык для Telegram Mini App (SvelteKit + Express). Покрывает Sentry error tracking, custom metrics, performance monitoring, uptime checks, Telegram alerting, structured logging с pino, Grafana dashboards. Используется для observability и быстрого реагирования на проблемы в production.

Detailed protocol for working with subagents in task execution, code audit, and troubleshooting. This skill should be used when launching subagents for implementation, code audit (phases 1-3), troubleshooting, or research tasks. Provides prompt templates, file formats, and strict rules to ensure subagents work within defined scope.

Organize all outputs from slash commands and subagents in assets/ directory by topics, date format, and slugs.

SEO audits, keyword research with ReviewWeb.site API (keyword ideas, difficulty, traffic, backlinks), on-page optimization, technical SEO, programmatic SEO (pSEO), JSON+LD schema, Google Search Console API integration. Activate for search optimization, meta tags, sitemap, Core Web Vitals. Used by seo-specialist, attraction-specialist, content-creator agents.

Use CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.

End-to-end food-record evaluation pipeline. Use when running VLM evals, CLIP food/non-food screening, CLIP→VLM batch extraction with resume, generating food calendars (month/week), or launching full background runs. Includes one-click pipeline runner and supporting scripts.

Shotgun codebase exploration - generates many commands, executes in parallel, extracts relevant context. Only 2 LLM calls.

Write tests to reach a coverage target. Uses 4 agents (analyzer, writer, checker, fixer) to iteratively generate and fix tests. Exits non-zero with explanation if production code needs changes.

Add a standard Purpose comment above each pure virtual method declaration in C++ interface header files (.h/.hpp). Use when asked to batch annotate interface classes or enforce a consistent comment block above `virtual ... = 0;` declarations.

Unity 项目 C# 编码标准和规范（包含 HybridCLR 热更新限制）

Hook-System Dokumentation und Patterns. Verwende bei hook, automation, PreToolUse, PostToolUse, SessionStart, event-driven

多 Agent 任務分解框架 - 將計劃轉化為可執行的 DAG 任務清單

Show help and documentation for inference-planz plugin

Intelligently commits and pushes git changes using conventional commit messages, automatically grouping files by logical task. Use when committing multiple features, fixes, or changes that should be separated into distinct commits.

Analyze ClickUp sprint metrics for H2 2025 (Sprint 43-50). Generates charts, calculates KPIs, and detects carryover tasks. Use when analyzing sprint performance, creating reports, or reviewing team velocity.

WatermelonDB models, observation patterns, and React integration. Use when writing or debugging model code, observers (findAndObserve, query.observe), or screens that display live-updating DB data.

Specialist in data quality validation frameworks—Great Expectations, dbt tests, data contracts. Builds comprehensive data quality gates into pipelines for reliability and trust.

Angular 21 development with modern best practices including signals, standalone components, reactive patterns, and accessibility. Use when creating Angular components, services, templates, or performing any Angular frontend development work. Covers TypeScript strict typing, signal-based state management, reactive forms, lazy loading, ng-icon setup, and Tailwind styling.

Angular standalone components, RxJS patterns, and reactive forms for frontend development. Use when: Building Angular components, services, guards, interceptors, or working with RxJS observables.

A self-reflexive verification skill that ensures agent compliance with LoFi Gate rules (Anti-Cheat, Intent Check).

Convert HTML files to Markdown format with intelligent preprocessing. Use when: (1) Converting single HTML file to Markdown, (2) Batch converting HTML files in a directory, (3) Processing saved web pages (SingleFile), (4) Converting documentation sites to Markdown.

Use when setting up a project to run with multi-worker coordinator, structuring features.json, creating PRD docs, or preparing for parallel feature implementation

在你准备声称“已完成/已修复/已通过”之前使用（尤其在 commit 或提 PR 前）：必须运行 verification 命令并核对输出；永远 Evidence before assertions。

在收到 code review 反馈时使用（在落地建议之前，尤其当反馈不清晰或技术上可疑时）：要求技术严谨与验证，拒绝表演式同意或盲目实现。

当你面对 2 个以上彼此独立的任务（无共享状态、无顺序依赖）时使用：为每个独立问题域派发一个 agent 并行推进。

Remove historical cruft from design docs. Use when docs accumulate outdated content, after major refactorings, or when preparing for releases.

Compare doc versions across git history. Use when reviewing changes, tracking evolution, or understanding modifications.

Comprehensive health audit for design documentation. Use when performing thorough quality checks, preparing for releases, or ensuring documentation compliance.

Set up Git hooks (Husky, lint-staged, commitlint) for commit-time code quality enforcement. Configures pre-commit, commit-msg, and pre-push hooks based on CLAUDE.md requirements.

Configure Claude Code PostToolUse hooks for real-time auto-linting. Automatically runs linters after Edit/Write operations for immediate code quality feedback.

Implements real observability infrastructure including Prometheus, Grafana, alerting rules, and dashboards. Use after observability-architect has designed the observability plan. Configures monitoring stacks, generates alert configurations, and creates operational dashboards.

Configure security scanning hooks for vulnerability detection and secrets scanning. Uses semgrep, bandit, and gitleaks for automated security analysis on code changes.

Use when user provides Jira issue URLs or mentions Jira tickets - fetches issue details and comments from Jira Cloud using local jira tool, outputs AI-optimized markdown for context gathering

Symbol kind (class, method, function, variable) - helps narrow results

Use this skill proactively when the user asks to rename a function, class, variable, or method. This performs safe semantic renaming across the entire codebase using LSP - never use find-and-replace for code refactoring.

USE THIS PROACTIVELY before modifying any code to check impact! Also use when debugging to trace where something is called from. Much more accurate than Grep - no false positives. Always use before refactoring or deleting code.

Recovery protocols when agent is stuck—escalate to new agent, migrate context to new session, or reset mid-conversation.

Converts video files to MP3 and transcribes them using OpenAI Whisper. Use when the user wants to get text from a video, transcribe a meeting recording, or extract speech from a video file.

Expert styling and UI/UX design system using the PurgeTSS toolkit for Titanium SDK. Use when Claude needs to: (1) Set up a new project with utility-first styling, (2) Implement complex Grid-based or responsive layouts, (3) Create declarative animations using the Animation component, (4) Automate assets like Icon Fonts or Color Palettes via CLI, (5) Configure advanced design rules in config.cjs, or (6) Work with arbitrary values, platform modifiers, and custom Ti Elements.

Comprehensive guide for the Alloy MVC framework for Titanium SDK. Use when working with Titanium Alloy applications - creating or modifying Alloy MVC components (models, views, controllers), implementing data binding with Backbone.js models and collections, styling with TSS (Titanium Style Sheets), working with widgets and components, using the Alloy CLI for code generation, implementing migrations and sync adapters, integrating PurgeTSS for utility-first styling. This skill contains the complete official Alloy documentation.

Titanium SDK UI/UX expert for layouts, ListViews, TableViews, events, animations, accessibility, scrolling, gestures, orientation, custom fonts, icons, splash screens, app structures, and platform-specific UI (Action Bar, Navigation Bar, iOS/Android components). Use when implementing complex UI, optimizing performance, handling platform differences, or configuring app icons and launch images.

Comprehensive Titanium SDK How-tos expert covering ALL official how-to guides. Use when Claude needs to implement: (1) Location & Maps with battery-efficient tracking, (2) Push/Local Notifications with proper lifecycle, (3) Media APIs (Camera, Gallery, Audio, Video, Images), (4) Remote data (HTTPClient, JSON, XML, uploads/downloads, sockets), (5) Local data (Filesystem, SQLite, Properties, Streams), (6) WebView integration with bidirectional communication, (7) Android platform features (Intents, Receivers, Services, Intent Filters), (8) iOS platform features (Background Services, Spotlight, Handoff, iCloud, Core Motion, WatchKit), (9) Extending Titanium (Hyperloop, native modules), or (10) Cross-platform patterns and persistence strategies.

This skill installs a complete Claude Code skill tracking system using hooks to automatically log skill usage, execution duration, token usage, and user prompts for later analysis. Use this skill when setting up activity tracking in a Claude Code project to identify patterns, monitor costs, and discover opportunities for new skills.

Installs and configures project infrastructure including MkDocs Material intelligent textbook templates, learning graph viewers, and skill tracking systems. Routes to the appropriate installation guide based on what the user needs to set up.

Create Frappe database patches and migrations. Use when creating migrations, patches, data transformations, schema changes, or when user mentions database migration, data update, or patch.

Push committed changes to remote repository. Runs pre-push validation including tests and brain sync check. Use after commits are ready for remote.

240+ prompt patterns for content analysis and transformation. USE WHEN fabric, extract wisdom, summarize, threat model.

Comprehensive research and content extraction. USE WHEN research, investigate, extract wisdom, analyze content. For OSINT use OSINT skill.

Prime agent context with roadmap, strategy, and codebase patterns. Run at session start.

Use when configuring cross-space access, sharing context between spaces, delegating tasks to other agents, or setting up agent-to-agent communication

Use when automating recurring tasks in spaces, setting up cron-like schedules, running agent prompts on a timer, or configuring periodic automation

Transform room photos into redesigned spaces using Pinterest inspiration images. Use when a user provides a photo of their room AND Pinterest/inspiration images, and wants to (1) visualize what their room could look like redesigned, (2) get a shopping list with real product links to manifest the design, or (3) receive interior design recommendations based on their style preferences. Creates visual mockups as React artifacts and searches the live web for purchasable products.