mcp-think
MCP server from nyanta012/mcp-think-server
claude mcp add --transport stdio nyanta012-mcp-think-server uv --directory C:\Users\{user}\mcp_server\think_server run server.pyHow to use
The Think server is a Python-based MCP server meant to run a local Think-enabled assistant inside the Think_Server directory. It is designed to be invoked via the UV runner (uv) and integrated with Claude Desktop through the claude_desktop_config.json file. When configured, Claude Desktop can route requests to this Think server, enabling you to interact with the Think model workflows implemented in server.py. To use it, ensure the Think server files are in place and that Claude Desktop points to the correct uv command with the appropriate working directory. Once running, clients can send prompts to the server just like other MCP endpoints and receive Think-based responses.
With this server, you gain access to the custom Think logic implemented in server.py, including any domain-specific prompt handling, memory, or orchestration routines embedded in that script. The integration with Claude Desktop means you can test and deploy Think capabilities directly from your GUI, using the same mcpServers.json configuration approach used by other MCP servers in your environment.
How to install
Prerequisites:
- Python 3.x installed on the host machine
- Access to the Think server files (think_server directory) containing server.py
- Claude Desktop or an MCP-capable environment if you plan to use the same mcpServers.json approach
Installation steps:
- Prepare Python environment
# Optional: create a virtual environment
python -m venv venv
source venv/bin/activate # on Windows use: venv\Scripts\activate
- Place Think server files
- Ensure the Think server directory exists at your chosen path, for example: C:\Users{user}\mcp_server\think_server
- The directory must contain server.py and any required dependencies
- Verify dependencies
pip install -r requirements.txt # if a requirements file exists in think_server
- Run the server with the UV runner
uv --directory C:\Users\{user}\mcp_server\think_server run server.py
- Connect via Claude Desktop
- Update claude_desktop_config.json with the provided mcpServers entry (see README snippet) so Claude Desktop can launch and interact with the Think server.
Additional notes
Tips and common issues:
- Ensure you replace {user} with your actual Windows username in the path.
- The Windows path in JSON must escape backslashes (e.g., C:\Users{user}...).
- The Think server script (server.py) must be present in the think_server directory; otherwise, uv will fail to start.
- If you modify server.py, restart the uv process to pick up changes.
- If Claude Desktop cannot reach the server, verify the directory path, permissions, and that Python/uv are correctly installed and on your PATH.
Related MCP Servers
mcp-vegalite
MCP server from isaacwasserman/mcp-vegalite-server
github-chat
A Model Context Protocol (MCP) for analyzing and querying GitHub repositories using the GitHub Chat API.
nautex
MCP server for guiding Coding Agents via end-to-end requirements to implementation plan pipeline
pagerduty
PagerDuty's official local MCP (Model Context Protocol) server which provides tools to interact with your PagerDuty account directly from your MCP-enabled client.
futu-stock
mcp server for futuniuniu stock
mcp -boilerplate
Boilerplate using one of the 'better' ways to build MCP Servers. Written using FastMCP