💬 Get started quickly - chat with your servers on our hosted version! Try mcp-use chat (beta) .
💡 Let developers easily connect any LLM to tools like web browsing, file operations, and more.
MCP.Pizza Chef: mcp-use
With about six lines of Python, mcp-use wires the model of your choosing to one or many MCP servers, such as a browser, your files, or a database, and lets it pick the right one for each request. You can block risky tools, run several servers at once, and stream results as they arrive. There is also a hosted chat if you would rather not install anything. Needs Python and your own model account key.
Coding
Web/Research
Use This MCP client To
Build a small assistant that browses the web for me Connect one agent to several of my tools at once Try the same task on different AI models Block tools I do not want an agent touching Chat with my MCP servers without writing any code
README
mcp-use FAQ
Do I need to write code?
For the full version yes, in Python. There is also a hosted chat you can try in your browser first.
Which AI models work with it?
Any model that supports tool calling, including the major providers and models you run locally.
Do I need my own model account key?
Yes. You add a key for whichever provider you choose.
Can I use this to give a model access to a browser?
Yes. Point it at a browser MCP server and the model can drive the pages itself.
Can it stop an agent from touching sensitive tools?
Yes. You can restrict which tools an agent is allowed to call, including file and network access.
Is it free?
The library is open source and free. You pay only for the model you use.