Enhanced Sequential Thinking MCP Server

View original on GitHub

This project implements an advanced Sequential Thinking server using the Model Context Protocol (MCP). It provides a structured and flexible approach to problem-solving and decision-making through a series of thought steps, incorporating stages, scoring, and tagging.

Features

🧠 Structured Problem Solving: Break down complex problems into defined stages
📊 Progress Tracking: Monitor thought sequences, branches, and revisions
🏷️ Thought Categorization: Tag and score thoughts for better organization
📈 Dynamic Adaptation: Adjust the thinking process as new insights emerge
📝 Summary Generation: Get an overview of the entire thinking process

Prerequisites

Python 3.11 or higher
UV package manager (Install Guide )

Project Structure


mcp-sequential-thinking/
├── mcp_sequential_thinking/
│   ├── server.py
│   └── __init__.py
├── README.md
└── pyproject.toml

Quick Start

Set Up Project


# Create and activate virtual environment
uv venv
.venv\Scripts\activate  # Windows
source .venv/bin/activate  # Unix
 
# Install package and dependencies
uv pip install -e .

Run the Server


cd mcp_sequential_thinking
uv run server.py

Claude Desktop Integration

Add to your Claude Desktop configuration (%APPDATA%\Claude\claude_desktop_config.json on Windows):


{
  "mcpServers": {
    "sequential-thinking": {
      "command": "uv",
      "args": [
        "--directory",
        "C:\\path\\to\\your\\mcp-sequential-thinking\\mcp_sequential_thinking",
        "run",
        "server.py"
      ]
    }
  }
}

API

The server exposes two main tools:

1. `sequential_thinking`

This tool processes individual thoughts in the sequential thinking process.

Parameters:

thought (str): The content of the current thought
thought_number (int): The sequence number of the current thought
total_thoughts (int): The total number of thoughts expected
next_thought_needed (bool): Whether another thought is needed
stage (str): The current stage of thinking (Problem Definition, Analysis, Ideation, Evaluation, Conclusion)
is_revision (bool, optional): Whether this revises previous thinking
revises_thought (int, optional): Which thought is being reconsidered
branch_from_thought (int, optional): Branching point thought number
branch_id (str, optional): Branch identifier
needs_more_thoughts (bool, optional): If more thoughts are needed
score (float, optional): Score for the thought (0.0 to 1.0)
tags (List[str], optional): List of tags for categorizing the thought

2. `get_thinking_summary`

This tool generates a summary of the entire thinking process.

Troubleshooting

Common issues:

Server Connection Issues
- Verify paths in claude_desktop_config.json
- Check Claude Desktop logs: %APPDATA%\Claude\logs
- Test manual server start

License

MIT License

Acknowledgments

Model Context Protocol framework
Claude Desktop team