Skip to content

mcp-use

Connect any LLM to MCP tools with a few lines of code

Visit official site

What mcp-use is

The library is described as an open-source MCP client that connects any large language model to MCP tools for building custom agents with tool access. It is stated to offer a simple API without a closed-source or application client.

What you can do with it

  • Create a first MCP-enabled agent in six lines of code
  • Use any LangChain LLM that supports tool calling, such as OpenAI and Anthropic
  • Connect directly to MCP servers running on a specific HTTP port
  • Run multiple MCP servers inside a single agent

Example uses

  • Browse the web with Playwright to look for restaurants in San Francisco
  • Check lodging options in Barcelona through Airbnb
  • Build 3D models and complex graphics in Blender

Getting started

  • Install the library and the required LLM provider
  • Set environment variables and add API keys
  • Create an MCPClient configuration that defines the required MCP servers
  • Instantiate the LLM and the MCPAgent

Pros & cons

✓ What we like

  • No closed-source or application client is needed
  • Dangerous tools such as file system or network access can be restricted

! What to watch out for

  • An API key must be set as an environment variable before use
  • HTTP connections require an MCP server running on a specific HTTP port

FAQ

How much code starts an agent?

The material states six lines of code create the first MCP-enabled agent.

Which LLMs are supported?

The feature list names any LangChain LLM with tool calling, such as OpenAI and Anthropic.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review