trackmcp
Back to directory
michaelneale

deepseek-r1-mcp

View on GitHub

a deepseek r1 MCP for reasoning/desigining with agents

2 stars JavaScriptAI & Machine Learning Updated Jul 2, 2025

Documentation

Deepseek r1 reasoning extension for locally running models

Aim: provide locally runnable technical agent system, using a combination of models and no remote models.

Uses deepseek r1 as a reasoning model, which can then be coupled with a tool calling model, for example in the case of goose agent below.

Deepseek r1 and other reasoning models often can't effectively do tool calling, but can be combined with other models which while not as good at reasoning, can do tool calling.

Below is a diagram of how it works with the "project goose" agent, which makes use of ollama with all local models:

image

This is a generic MCP server so should work with any agent system, as long as ollama is running, and can run entirely locally.

Usage:

sh
npx -y deepseek-reasoner-mcp

This can be used in any agent system.

Frequently asked questions

What is deepseek-r1-mcp?

deepseek-r1-mcp is a deepseek r1 MCP for reasoning/desigining with agents

How do I install deepseek-r1-mcp?

Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.

Is deepseek-r1-mcp open source?

Yes — it is hosted on GitHub at https://github.com/michaelneale/deepseek-r1-mcp and has 2 stars.

Related MCP tools

Run your own MCP server? See who uses it and what to fix.

Measure it with TrackMCP