MCP Reasoner

A reasoning implementation for Claude Desktop, enhancing problem-solving with Beam Search and Monte Carlo Tree Search (MCTS).

Current Version:

v2.0.0

What's New:

Added 2 Experimental Reasoning Algorithms:

- `mcts-002-alpha`: A* Search with an early Policy Simulation Layer and Adaptive Exploration & Outcome Based Reasoning Simulators (alpha).

- `mcts-002alt-alpha`: Bidirectional Search with an early Policy Simulation Layer and Adaptive Exploration & Outcome Based Reasoning Simulators (alpha).

Previous Versions:

v1.1.0

Added model control over search parameters: beamWidth (1-10) and numSimulations (1-150).

Features

• Beam search for straightforward tasks.
• MCTS for complex problems, including alpha variations.
• Tracks reasoning path quality and maps Claude's thought processes.
• Analyzes the reasoning process and follows the MCP protocol.

Installation

git clone https://github.com/frgmt0/mcp-reasoner.git 
cd mcp-reasoner
npm install
npm run build

Configuration

Add to Claude Desktop config:

{
  "mcpServers": {
    "mcp-reasoner": {
      "command": "node",
      "args": ["path/to/mcp-reasoner/dist/index.js"],
    }
  }
}

Testing

[More Testing Coming Soon]

Benchmarks

[Benchmarking will be added soon]

Key Benchmarks to test against: MATH500, GPQA-Diamond, GMSK8, Polyglot, SWE-Bench.

License

MIT License.