Skip to main content
AgentsUse

Directory

MCP Servers Directory

Model Context Protocol servers connect an agent client to files, repositories, browsers, and hosted APIs. Every page here shows the transport, a placeholder configuration checked against the official source, the environment variable names it needs, the tools it exposes, and the security scope you are granting before you paste anything into a client config.

16 verified server setups. How we verify: read the methodology.

stdioFilesystem MCP ServerNode.js MCP server for reading, writing, searching and managing files inside explicitly allowed directories. By Model Context Protocol. 4 verified tools.stdiomcp-server-gitPython MCP server that lets a model inspect and change a local Git repository through a fixed set of Git tools. By Model Context Protocol. 4 verified tools.stdioFetch MCP ServerPython MCP server that fetches a web page and returns its content as markdown, in chunks when needed. By Model Context Protocol. 4 verified tools.stdioKnowledge Graph Memory ServerNode.js MCP server that stores persistent memory as a local knowledge graph of entities, relations and observations. By Model Context Protocol. 4 verified tools.stdioSequential Thinking MCP ServerNode.js MCP server with one tool for step by step problem solving that supports revisions and branching. By Model Context Protocol. 4 verified tools.stdioTime MCP ServerPython MCP server that returns the current time in an IANA timezone and converts times between timezones. By Model Context Protocol. 3 verified tools.stdio / sse / streamable-httpEverything MCP ServerReference test server that exercises MCP tools, resources, prompts, sampling and multiple transports for client builders. By Model Context Protocol. 5 verified tools.stdio / ssePlaywright MCP ServerMicrosoft MCP server that automates a real browser through Playwright using accessibility snapshots instead of screenshots. By Microsoft. 6 verified tools. Related tool profile: Playwright.stdioChrome DevTools MCP ServerAn MCP server that lets coding agents control and inspect a live Chrome browser using Chrome DevTools for automation, debugging, and performance analysis. By ChromeDevTools (Google Chrome DevTools team). 3 verified tools.streamable-http / stdioGitHub MCP ServerGitHub official MCP server that connects AI tools directly to GitHub to read repositories and code, manage issues and pull requests, analyze code, and automate workflows. By GitHub. 4 verified tools.streamable-httpContext7 MCP ServerContext7 pulls up-to-date, version-specific documentation and code examples from the source and places them directly into the prompt for LLMs and AI code editors. By Upstash. 4 verified tools.stdio / streamable-http / sseFirecrawl MCP ServerOfficial Firecrawl MCP server that adds web scraping, search, crawling, and page interaction to Cursor, Claude, and other LLM clients for clean, agent-ready context. By Firecrawl. 5 verified tools. Related tool profile: Firecrawl.streamable-http / stdioTavily MCP ServerProduction ready MCP server with real-time search, extract, map, and crawl tools powered by Tavily. By Tavily. 4 verified tools. Related tool profile: Tavily.streamable-httpExa MCP ServerConnect AI agents to Exa for web search, content fetching, and multi-step research. By Exa Labs. 4 verified tools. Related tool profile: Exa.streamable-httpSupabase MCP ServerConnect your Supabase projects to Cursor, Claude, Windsurf, and other AI assistants to manage tables, fetch config, and query data. By Supabase. 4 verified tools.stdio / streamable-httpNotion MCP ServerOfficial Notion MCP Server implementing an MCP server for the Notion API, with a separate hosted Remote Notion MCP recommended by Notion for the best experience. By Notion (makenotion). 7 verified tools.

Before you paste a config

A local MCP server runs with your user permissions. A filesystem server can read what its paths allow, a shell or database server can change real data, and a remote server sends your prompts to a third party. Start with read-only credentials and a test directory, keep keys in environment variables rather than shared files, and re-read the security notes on each page when you widen scope.