MCP Think Tank 🧠

A next-generation knowledge graph and reasoning tool built with FastMCP V2. This is a complete Python rewrite of the original TypeScript version with enhanced capabilities.

Features

• Advanced Knowledge Graph - Store and retrieve information with semantic search capabilities
• Structured Reasoning - Think and reflect on complex problems with chain-of-thought
• Task Management - Plan and organize multi-step workflows
• Cross-Tool Intelligence - Orchestrated interactions between memory, reasoning, and tasks
• File System Awareness - Index and search code in your projects

Installation

Prerequisites

• Python 3.12+
• uv for dependency management

Install with uv

# Clone the repository
git clone https://github.com/flight505/mcp-think-tank-py.git
cd mcp-think-tank-py

# Create a virtual environment and install dependencies
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e .

Install with Claude Desktop

# Install directly for Claude Desktop
fastmcp install src/server.py --name "MCP Think Tank"

Quick Start

1. Run in development mode

fastmcp dev src/server.py

2. Use with Claude Desktop

After installing with fastmcp install, the MCP Think Tank will be available as a tool in Claude Desktop.

Usage

Knowledge Graph

# Create entities
create_entities(entities=[{"name": "FastMCP", "entityType": "Technology", "observations": ["A Pythonic MCP framework"]}])

# Create relations
create_relations(relations=[{"from": "FastMCP", "to": "Python", "relationType": "built with"}])

# Search for entities
search_nodes(query="FastMCP")

Structured Reasoning

# Think about a problem
think(structured_reasoning="Let me analyze how to approach this problem step by step...", store_in_memory=True)

Task Management

# Create tasks from requirements
create_tasks(prd_text="Build a feature that integrates with our database and provides a REST API")

# List current tasks
list_tasks()

Architecture

MCP Think Tank is built with a modular architecture:

• Core Server - FastMCP-based server implementation
• Orchestrator - Coordinates interactions between tools
• Tools - Memory, Think, and Task functionality
• Storage - JSONL-based persistence with semantic search

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• FastMCP for the powerful MCP server framework
• Model Context Protocol for standardizing LLM context provision