think

Think MCP Server is a Python-based server that implements the MCP protocol and provides tools for prompt management, resource handling, and utility calls. It exposes capabilities to manage prompt templates, access and organize resources, and invoke analysis tools for content and article analysis. Once running, you can interact with its interfaces through the MCP protocol and configure behavior via environment variables stored in a .env file. Typical usage involves starting the server with the uvx runner and using the built-in prompts and resources APIs to fetch, read, or modify data, as well as invoking the available analysis tools provided by the server. The server is designed to be cross-platform and self-managing configuration on first run, creating necessary directories and files under a user’s home directory.

To operate it in Windsurf or similar MCP clients, configure the server entry to point to the uvx runner and the think-mcp-server executable. You can list prompts, fetch specific prompts, read resources, and call content or article analysis tools through the MCP endpoints exposed by the server.

Prerequisites:

• Python environment (3.8+ recommended)
• Poetry or pip for dependencies
• Internet access to install dependencies

Option A: Install from PyPI (recommended for end users)

1. Install the package: pip install think-mcp-server
2. Run the server directly using uvx (if you have uvx installed): uvx think-mcp-server

Option B: Install from source (developers)

1. Clone the repository: git clone https://github.com/thinkthinking/think-mcp-server.git
2. Change into the project directory: cd think-mcp-server
3. Install in editable mode (development): pip install -e .
4. Run the server with uvx: uvx think-mcp-server

Prerequisites for running:

• Ensure uvx is installed and available in PATH if you plan to run via uvx
• Ensure Python virtual environment is active to avoid dependency conflicts
• Optionally set up a .env file at ~/.think-mcp-server/config/.env for environment variables

Billing note: No special build steps are required; the project uses standard Python packaging.

Environment and configuration tips:

• On first run, the server will create configuration directories at ~/.think-mcp-server/ including config/, prompts/, and resources/.
• Environment variables (prompts_path, resources_path, article_base_path, max_tokens_limit, content_analyzer_prompt_path) can be customized in ~/.think-mcp-server/config/.env.
• The prompts directory holds prompt templates; resources holds resources used by tools; article_base_path configures article analysis inputs. Adjust paths to fit your local filesystem.
• If you encounter issues, check logs under the server's log location (configured in .env or your default logging setup) for troubleshooting hints.
• This server supports a cross-platform setup (macOS, Windows, Linux). Ensure file permissions allow the server to write to its config and data directories.
• The server exposes two main tooling capabilities: content analysis (analyzing file contents to extract key information) and article analysis (analyzing article structure and content). These are available via the MCP interface as handle_read_resource, handle_list_prompts, handle_get_prompt, and specific tool invocation endpoints.