
Best Claude Skills, MCP Servers & AI Agents
The largest directory of Skills, Plugins, Agents, MCP Servers, and Prompts for Claude Code, Cursor & Windsurf. Open source, verified from GitHub, updated hourly.
Trending This Week
Openclaw MCP Server
Openclaw is a multi-channel AI gateway MCP server that enables extensible messaging integrations across AI platforms. It benefits developers building personal AI assistants who want to own their data and integrate multiple communication channels.
gh-issues
An AI-powered GitHub issue orchestrator that automatically fetches issues, spawns sub-agents to implement fixes, opens PRs, and monitors review feedback. Ideal for developers wanting to automate bug triage and fix deployment workflows.
finishing-a-development-branch
Guides developers through the final stages of completing a development branch by verifying tests pass and presenting structured options for merging, creating PRs, or cleanup. Useful for developers who want a systematic approach to integrating completed work.
subagent-driven-development
A skill booster that structures implementation execution by dispatching independent subagents for each task with two-stage review (spec compliance, then code quality), designed for developers managing multi-task implementation plans in a single session.
using-superpowers
This booster enforces mandatory skill tool invocation at the start of conversations to establish available capabilities. It's designed for users who want to ensure AI assistants always discover and declare applicable skills before responding.
test-driven-development
Guides AI coding assistants to follow Test-Driven Development principles by writing failing tests before implementation code. Useful for developers who want to maintain code quality and ensure comprehensive test coverage.
using-git-worktrees
Helps developers create isolated git worktrees for parallel feature development without disrupting their current workspace. Essential for teams managing multiple branches simultaneously.
brainstorming
A collaborative design-first booster that guides developers through brainstorming and specification before implementation, preventing premature coding and misaligned requirements.
Browse by Category
Development Tools
Code review, linting, testing, and dev workflow
Writing & Productivity
Documentation, content creation, and writing assistance
Data & Research
Databases, analytics, APIs, and data pipelines
DevOps & Infrastructure
Deploy, monitor, and scale cloud infrastructure
Browser & Automation
Web scraping, UI testing, and browser automation
Security
Auth, encryption, and vulnerability scanning
Design & Media
UI/UX, image, video, and creative tools
AI & Machine Learning
LLMs, embeddings, RAG, and AI model tools
Recently Added
codeflow
CodeFlow (braedonsaunders, MIT) turns any GitHub repository, local folder, PR, or markdown/Obsidian vault into an interactive architecture map — entirely in the browser. It is a single
runcomfy-cli
One binary, one auth, every RunComfy model. Install once, sign in once, then call any text-to-image, video, edit, lip-sync, face-swap, or LoRA-training endpoint with . This skill is the foundation every other skill builds on. runcomfy.com · CLI docs · All models A standalone curl-pipe installer als
pwn-www-twitter
This plugin supports twitter.com actions. Call from when the task needs this module. Do not reimplement it in shell.
ideogram-ci-integration
Make Ideogram integration quality reproducible without granting untrusted changes a paid credential. Use fixtures and schema checks for every pull request, then isolate any live smoke test behind trusted-branch, synthetic-content, spend, and concurrency controls. The first-party OpenAPI is machine-r
diagnose
Systematic diagnosis of test failures in the Morphia project. Run the test, read the full error, trace the root cause through source and generated code, fix minimally, and verify. 1. Run the failing test 2. Read the full error - stack trace, error message, line numbers. Don't skip anything.
humanize-writing
You are an expert editor who specializes in detecting and removing AI writing patterns. Your job is to take content that reads like it was generated by a language model and rewrite it so it sounds like a knowledgeable human wrote it on the first try. AI writing has a recognizable smell. It's not abo
consulting-chart
PNG 렌더 후 반드시 5단계 검증을 실행한다. 하나라도 FAIL이면 수정 후 재렌더 (최대 3회). Read로 PNG를 열어 모든 텍스트가 눈으로 읽히는지 확인: McKinsey/Bain/BCG consulting report standard:
teacher
自动检测用户第一条消息的语言。整个会话过程中始终使用该语言。如果用户中途切换语言,跟随用户切换。 你是一位耐心、通俗、因材施教的 AI 老师。 你的核心使命只有一个:确保用户真正理解给定的资料内容,而不是死记硬背。
Find your perfect
AI stack
Not sure where to start? Get a curated combination of Boosters matched to your workflow and tools.
Try the Stack Builder