Browse Model Context Protocol (MCP) servers and integrations.
Model Context Protocol (MCP) servers are standardized interfaces that allow large language models and AI agents to securely interact with external systems, tools, and data stores. Instead of relying on bespoke integrations, an MCP server exposes defined tools, resources, and prompts over a unified protocol. This architecture enables AI coding assistants, desktop agents, and autonomous workflows to query databases, manipulate local files, execute terminal commands, and call external APIs with fine-grained control.
This directory catalogs 1186 active MCP servers structured across 20 functional domains. Developers can browse servers organized by target environment and capability, including Databases & Data Stores for querying relational and vector backends, Developer Tools & Code Intelligence for IDE enhancements, and DevOps, CI/CD & Version Control for deployment automation. Additional domains include Cloud & Infrastructure, Security & Compliance, Browser & Web Automation, AI Memory & Context, and Files, Documents & PDFs. Each category groups servers designed to grant models access to specific operational layers without requiring custom middleware.
Selecting the right MCP server depends on your client environment, transport protocol, and execution runtime. First, confirm compatibility with your AI client, such as Claude Desktop, Cursor, Windsurf, or custom agent runtimes built on the MCP SDK. Second, inspect the server transport type; most run locally over standard input/output (stdio), while others operate remotely via Server-Sent Events (SSE) or WebSockets. Finally, evaluate the required permission scopes and authentication mechanisms. Production workflows benefit from servers that enforce read-only modes or explicit confirmation prompts prior to executing destructive actions on your infrastructure.
| MCP server | What it connects to | Type | Repository |
|---|---|---|---|
| Qdrant MCP Server | Qdrant MCP Server is an MCP server that provides semantic code search capabilities across software repositories using the Qdrant vector | Open Source | https://github.com/steiner385/qdrant-mcp-server |
| YouTube Data MCP | The YouTube Data MCP acts as a smart bridge between AI models and the vast library of content on YouTube. | Open Source | https://github.com/kirbah/mcp-youtube |
| YouTube | The YouTube MCP server acts as a bridge between video content and AI assistants like Claude Desktop, allowing users to | Unknown | https://github.com/bendelpino/youtube-mcp |
| Memory-Plus | Memory-Plus is an MCP server that provides a lightweight, local Retrieval-Augmented Generation (RAG) memory store for AI agents across different | Open Source | https://github.com/Yuchen20/Memory-Plus |
| Mealie | Mealie MCP Server is an MCP server that connects AI assistants like Claude Desktop directly to a running Mealie recipe | Open Source | https://github.com/rldiao/mealie-mcp-server |
| O'RLY Book Cover Generator | O'RLY Book Cover Generator is an MCP server that produces O'RLY parody book covers and displays the generated images directly | Unknown | https://github.com/charleshberlin/O-RLY-Book-Generator |
| Turso Cloud | Turso Cloud is an MCP server that integrates Turso databases with large language models through a secure two-level authentication model. | Unknown | https://github.com/spences10/mcp-turso-cloud |
| Tushare MCP | Tushare MCP is an MCP server that connects AI assistants like Claude Desktop and Cursor to financial market data from | Unknown | https://github.com/buuzzy/tushare_MCP |
| Peekaboo MCP | Peekaboo MCP is an MCP server that provides strictly read-only access to a local project filesystem. Built for software engineers | Unknown | https://github.com/davstr1/peekabooMCP |
| Slack MCP Client in Go | Slack MCP Client in Go is an MCP client and Slack bot bridge that connects Slack conversations directly to Model | Open Source | https://github.com/tuannvm/slack-mcp-client |
An MCP server is an application that implements the Model Context Protocol to expose data, tools, and prompts to AI assistants. It standardizes how models read external context and trigger actions in external systems.
Clients supporting MCP include Claude Desktop, Cursor, Windsurf, Continue, and various open-source autonomous agent frameworks that implement the Model Context Protocol client specifications.
The majority of MCP servers are open-source and free to run locally. However, some servers interface with third-party commercial APIs, cloud providers, or proprietary platforms that require paid subscriptions or API keys.
Developers can submit a server by providing its repository link, documentation, supported transport protocols (stdio or SSE), license type, and category classification through the FastPedia submission form.
Listings are ranked using objective criteria, including repository maintenance activity, implementation completeness, documentation quality, community usage signals, and protocol compliance.
MCP servers execute locally with the permissions of the host environment. Users should inspect server source code, verify required file system and network privileges, and configure client-side approval prompts before granting execution rights.