Explainer
What is Model Context Protocol (MCP)?
Model Context Protocol — MCP for short — is an open standard for connecting AI models to the outside world. It was created by Anthropic in late 2024 and quickly became the de-facto way to give AI agents access to tools, files, APIs, databases, and live data.
Before MCP, every AI client had to ship its own custom plugin system. OpenAI had functions, Anthropic had tools, every other vendor had something different. The result: tool builders had to re-implement the same integration four times, and users could not move their workflow between models. MCP fixes that with a single client–server protocol over JSON-RPC.
Why it matters
MCP turns "AI assistants" into "AI agents." A model that can query your Postgres database, file GitHub issues, and post to Slack — all using one common protocol — is genuinely useful in a way chat-only models are not. And because MCP is open, the ecosystem grows much faster than any single vendor could grow it alone.
How it works
An MCP client (like Claude Desktop, Zed, or Cursor) launches one or more MCP servers as subprocesses, communicating over stdio or HTTP. Each server exposes a list of tools, resources, and prompts. The model decides when to call a tool, the client routes the call, and the server runs the work and returns the result.
How to get started
- Install an MCP-compatible client (Claude Desktop is the easiest starting point).
- Pick an MCP server from this directory.
- Copy its one-line install command into your client's MCP config.
- Restart the client — your AI model now has access to the new capability.
From there it's additive: stack as many servers as you want. The Filesystem MCP, Memory MCP, and GitHub MCP together turn Claude into a full development assistant in about three minutes.
Frequently Asked Questions
What is Model Context Protocol?+
Model Context Protocol (MCP) is an open standard for connecting AI models to external tools, data sources and APIs. It defines a common way for AI clients (like Claude Desktop) to discover and call capabilities exposed by MCP servers.
Who created MCP?+
MCP was created by Anthropic and released as an open standard in late 2024. It is now implemented by a growing ecosystem of clients and servers across companies and open-source projects.
What can you do with MCP servers?+
MCP servers let you give AI access to anything: databases, file systems, GitHub repos, Slack workspaces, Notion pages, search engines, browser automation, payments — the list grows daily.
How do I use MCP with Claude?+
Install Claude Desktop, add an MCP server to its config file, and restart. Most servers in this directory ship with a one-line npx install command — just paste it into your Claude Desktop MCP config.
Are MCP tools free?+
Most MCP servers are free and open-source. Some commercial integrations charge for usage (e.g. search APIs, image generation), but the protocol itself is free.
What is the difference between MCP tools and MCP resources?+
Tools are actions the model can call (read a file, query a DB). Resources are data the model can read but not invoke (a list of issues, a document). Some MCP servers expose both.