Discover the best of the
Claude ecosystem
A curated, real-time directory of the top GitHub repositories for Claude Code, skills, subagents, MCP servers, hooks, and templates, ranked by stars and activity.
Connect Claude to the tools you use
Official hosted MCP servers, verified in each vendor's own documentation.
Top 3 this week
The hottest Claude repositories, ranked by stars × velocity × recency.
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
The agent that grows with you
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
Chasing the podium
Full leaderboardProduction-ready platform for agentic workflow development.
The agent engineering platform.
Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)
An open-source AI agent that brings the power of Gemini directly into your terminal.
A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI & Hermes Agent. Only official website: ccswitch.io
A collection of MCP servers.
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
What do you want to build?
Filter the ecosystem by use case: find tools for creative, dev, data, marketing and more.
The giants
Heavyweights of the Claude & Model Context Protocol universe.
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
The agent that grows with you
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发
Production-ready platform for agentic workflow development.
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)
The agent engineering platform.
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
Fresh commits
Active maintainers pushing in the last few days.
QA Skills Directory QA Skills is a curated directory of testing-specific skills for AI coding agents (Claude Code, Cursor, Copilot, etc.).
Open-source control plane for your AI agents. Connect tools, hire agents, track every token and dollar
Zero, your trustworthy AI teammate for real work.
Open-source local workbench for multi-agent software development.
Sub2API is an open-source relay platform that unifies Claude, OpenAI, Gemini, and Antigravity subscriptions into a single endpoint. It supports account sharing and cost-sharing, with seamless native tool compatibility.
An Obsidian plugin that embeds Claude Code/Codex as an AI collaborator in your vault
🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.
A Go framework for building production agent systems with graph workflows, tools, memory, A2A, AG-UI, MCP, evaluation, and observability.
The Agent Harness for AI-Human Collaboration, inspired by the AI-DLC (AI-Driven Development Lifecycle)
Fresh arrivals in the ecosystem
Repos GitHub saw born in the last 30 days that our daily sync just picked up for the directory.
Legal discovery for your AI, find lawsuits by name, CPF, CNPJ or case number and build a consolidated X-ray (cases, publications, sanctions, municipal mentions and case law). Works instantly, no login. An alternative to services like Escavador and Ju
MCP server manifest for PurchaseIQ — pre-purchase trust intelligence for AI agents
Aggregates bank data (balances, bills and statements) via Brazil's Open Finance. Each institution becomes a connection; multiple banks per user are stored in credentials.connections[]. Read-only.
Aggregates bank data (balances, bills and statements) via Brazil's Open Finance. Each institution becomes a connection; multiple banks per user are stored in credentials.connections[]. Read-only.
Aggregates bank data (balances, bills and statements) via Brazil's Open Finance. Each institution becomes a connection; multiple banks per user are stored in credentials.connections[]. Read-only.
Aggregates bank data (balances, bills and statements) via Brazil's Open Finance. Each institution becomes a connection; multiple banks per user are stored in credentials.connections[]. Read-only.
Aggregates bank data (balances, bills and statements) via Brazil's Open Finance. Each institution becomes a connection; multiple banks per user are stored in credentials.connections[]. Read-only.
Aggregates bank data (balances, bills and statements) via Brazil's Open Finance. Each institution becomes a connection; multiple banks per user are stored in credentials.connections[]. Read-only.
Compare the Claude models
When to pick Opus, Sonnet or Haiku based on context, cost and speed. Tap a metric to see how they stack up.
The most capable tier, built for the hardest long-horizon agentic and reasoning tasks.
Deep coding, complex agents, and advanced reasoning at a lower cost than Fable.
The ideal balance of speed and intelligence for most production workloads.
Fast and cost-efficient responses at scale for high-volume, latency-sensitive applications.
1,936 servers audited: 29% flagged, 11% unlicensed, 20% unmaintained.
Live analysis and news
What our editorial agents are writing and what's brewing across Anthropic, GitHub, Hacker News and Reddit, all in one place.
An editorial directory of the Claude AI ecosystem
ClaudeWave indexes 4,241 public GitHub repositories related to Claude AI, Skills, Subagents, MCP Servers, Hooks, Plugins, Templates, Tools and Awesome Lists. 9,700,352 total stars across 8 categories and 19 use cases. Refreshed daily by an automated pipeline that combines GitHub Search API, curated awesome-lists and GitHub Trending.
Every repo runs through our security agent (Claude Haiku 4.5) which assigns a Trust Score 0-100 with transparent criteria, passed checks, detected flags, last scan timestamp. It's the quality layer that's missing when you only look at stars: a 5K-star repo that's archived with no commits in 18 months gets flagged accordingly.
We maintain it openly from ElephantPink, an AI engineering team that ships custom Claude integrations, MCP servers, agent skills, RAG pipelines and AI workflow automation for product companies. ClaudeWave is not affiliated with Anthropic, we use the name 'Claude' descriptively. If you've built something in the ecosystem, submit it: we review every submission within 48h.
- 1Daily pipeline crawls GitHub Search, awesome-lists and Trending to discover new repos.
- 2Heat score combines stars, velocity (stars/week) and recency of the last push.
- 3Automatic Trust Score audits each repo for quality signals and risk markers.
Questions, answered
What is ClaudeWave?+
ClaudeWave is a free, community-built directory of the Claude AI ecosystem. We index over 3,955 GitHub repositories across 8 categories (skills, subagents, MCP servers, hooks, plugins, templates, tools, and awesome-lists), ranked by stars, velocity and activity. The catalog is refreshed daily.
Is ClaudeWave affiliated with Anthropic?+
No. ClaudeWave is an independent project built and maintained by ElephantPink, an AI engineering team that ships custom Claude integrations and MCP servers. We are not affiliated with, endorsed by, or sponsored by Anthropic PBC. "Claude" is used descriptively under nominative fair use.
How do I install an MCP server from ClaudeWave?+
Click any MCP server page. You will see an Install block with a ready-to-paste claude_desktop_config.json snippet. Copy it, merge into your existing config, replace placeholders with your API keys, and restart Claude Desktop. Takes about 30 seconds.
What is the Stack Builder?+
Select multiple MCP servers with checkboxes, and ClaudeWave generates a single combined claude_desktop_config.json with all of them configured. Copy or download. Ideal to set up your full Claude toolkit in 30 seconds.
How does the ranking work?+
We compute a heat score combining total stars, star velocity (growth per week), and recency of commits. Tiers are Legendary (top 5), Epic (6-20), Rare (21-50), Common (51+). Rankings refresh daily.
Can I submit my own repo?+
Yes, go to /submit. We review every submission within 48 hours. You can also just wait; our sync pipeline auto-discovers tagged repos (use claude-skill, mcp-server, claude-code, etc. as GitHub topics).
Is ClaudeWave free?+
Yes, 100%. No paywall, no sign-up required. Optional newsletter and Discord for updates.
What categories does ClaudeWave cover?+
Skills, Subagents, MCP Servers, Hooks, Plugins, Templates, Tools, and Awesome Lists. We also tag every repo with up to 3 use cases (Creative, Productivity, Dev Tools, Data, Marketing, and 14 more).
ClaudeWave Weekly
Every Monday: the 10 hottest new Claude repos, hidden gems, and ecosystem updates. Free, no spam, one-click unsubscribe.
Real-time chat with MCP server maintainers, Claude Code power users, and developers shipping on the Claude ecosystem. Dedicated channels for each category, weekly AMAs, and office hours.
Built something for Claude?
Submit your skill, MCP server, agent, or template. Get featured, earn a tier, and reach thousands of developers building on Claude.
Submit your repo