The open directory of Model Context Protocol servers — browse, search, and integrate tools into your AI workflows.
The complete platform for discovering, evaluating, and deploying MCP servers into production AI workflows.
Browse a curated directory of official and community-built MCP servers across databases, developer tools, cloud services, and more. Filter by language, transport type, and category.
Every server comes with ready-to-use install commands and JSON config examples. Copy and paste directly into your Claude Desktop or AI client config.
Dead and unreachable servers are automatically removed from the directory. The install configs you copy here point to servers that are actually up and running.
Each server entry lists every tool it exposes — from read_file to kubernetes_deploy — so you know exactly what capabilities you're adding to your AI. Complete with transport type, version history, and configuration examples.
Find the right integration for every part of your stack.
Hand-picked, production-ready integrations
Real-time infrastructure monitoring with metrics, logs, alerts, and ML-based anomaly detection.
Draw and visually collaborate with your agents on tldraw's canvas.
MCP server for Chrome DevTools
Validate oh-my-posh configurations and segment snippets against the official schema.
Search and discover Agent Skills from the skills.sh registry. Powered by HAPI MCP server.
Publish and discover MCP servers via the official MCP Registry. Powered by HAPI MCP server.
Fresh releases and recent updates
Everything you need to know about MCP servers and MCP Find.
MCP Find is an open-source directory of Model Context Protocol servers. It helps developers discover, evaluate, and install MCP servers into AI clients like Claude Desktop, Cursor, VS Code, and Windsurf.
MCP is an open standard created by Anthropic that lets AI assistants connect to external tools and data sources. Think of it like a USB port for AI — any MCP-compatible client can use any MCP server, regardless of who built it.
Find a server on MCP Find, copy the JSON config snippet from its listing page, and paste it into your AI client's configuration file. Most servers work with a single command or config block — no complex setup required.
The vast majority of MCP servers listed here are open source and free. Some connect to paid third-party services (like cloud providers or SaaS APIs), but the MCP server itself is typically free.
Claude Desktop, Claude Code, Cursor, VS Code (via Copilot), Windsurf, and a growing number of AI tools support MCP. Any client that implements the protocol can connect to the servers listed here.
Click "Submit a Server" in the navigation bar and fill out the form with your server's package name and GitHub URL. Submissions are reviewed and typically appear in the directory within a few days.
Guides, tutorials, and insights on MCP
PayPal splits its MCP support across two different repos, one naming OpenAI and one naming Claude. Stripe ships a single server that names all three major model providers plus a hosted OAuth mode. That split shapes which one fits your stack before you write a line of config.
GitHub ships its MCP server as its own repo you can star, fork, and read. GitLab built its server straight into the product, so there's no separate codebase to point to at all. If you're picking a git host with your AI agent in mind, that difference changes more than you'd think.
Intercom's engineering team maintains its own MCP server, six tools, hosted only, US workspaces only. Zendesk has no confirmed first-party equivalent, just a handful of community repos with different tool coverage and different auth models. Both can work well, but they're two very different kinds of bet.
Share it with the community. MCP Find makes it easy for developers to discover your integration and start using it in minutes.