MindMesh MCP Server

What is MindMesh MCP Server?

MindMesh MCP Server is a Model Context Protocol (MCP) server that creates a quantum-inspired swarm of Claude 3.7 Sonnet instances with field coherence optimization, enabling enriched reasoning through multiple specialized LLM instances that work together with emergent properties.

How to use MindMesh MCP Server?

Clone the repository, install dependencies with npm install, create a .env file with your API keys, then build and start with npm run build and npm start. For development with auto-reload, use npm run dev. Connect any MCP client (e.g., Claude Desktop, Cursor IDE, Cline VSCode extension) and invoke the main tool reason_with_swarm with a prompt to process it through the swarm.

Key features of MindMesh MCP Server

• Quantum-inspired field computing for coherence between instances
• WebContainer integration for sandboxed execution environment
• PGLite with vector storage for efficient state management
• Multiple specialized Claude instances for enriched reasoning
• Coherence optimization to select most consistent outputs
• Optional 128k token extended thinking capability

Use cases of MindMesh MCP Server

• Complex reasoning tasks requiring multiple perspectives
• Research and analysis of interdisciplinary topics
• Decision support with coherence validation across responses
• Enhanced query processing through specialized LLM collaboration

FAQ from MindMesh MCP Server

What are the prerequisites for MindMesh MCP Server?

Node.js 18+ and an Anthropic API key with access to Claude 3.7 Sonnet. A VoyageAI API key is optional but recommended for better embeddings.

How many Claude instances does MindMesh MCP Server use by default?

The default number is 8, configurable via the CLAUDE_INSTANCES environment variable.

Is a VoyageAI API key required?

No, it is optional. When not provided, the server falls back to a simpler deterministic embedding method.

What transport methods does MindMesh MCP Server support?

It supports HTTP by default (port 3000) and can also use stdio by setting STDIO_TRANSPORT=true in the .env file.

Where does MindMesh MCP Server store data?

It uses a PGLite database. The path is configurable via DB_PATH, defaulting to "idb://mindmesh.db".