mind-mcp 0.2.0

MIND Engine — Graph physics, traversal, client

mind-mcp

Mind Protocol MCP server and runtime for AI agents. Graph physics, traversal, structured dialogues.

Quick Start

# Clone and install
git clone https://github.com/mind-protocol/mind-mcp.git
cd mind-mcp
pip install -e .

# Initialize a project (defaults to FalkorDB)
mind init

This creates .runtime/ with:

• Protocol docs (PRINCIPLES.md, FRAMEWORK.md)
• Agent definitions, skills, procedures
• Python runtime for physics, graph, traversal
• Database config (graph name defaults to repo name)

Local Runtime

After mind init, projects can run mind locally without pip install:

PYTHONPATH=".mind:$PYTHONPATH" python3 my_script.py

# my_script.py
from mind.physics.constants import DECAY_RATE
from mind.connectome import ConnectomeRunner
from mind.infrastructure.database.factory import get_database_adapter

adapter = get_database_adapter()

CLI Commands

mind init [--database falkordb|neo4j]  # Initialize .runtime/ with runtime
mind status                             # Show status and modules
mind upgrade                            # Check for updates

Database Backends

Graph name defaults to repo name (e.g., my-project → my_project).

FalkorDB (default, local)

mind init

# Start FalkorDB
docker run -p 6379:6379 falkordb/falkordb

Override graph name in .env:

FALKORDB_GRAPH=custom_name

Neo4j (cloud or local)

mind init --database neo4j

Configure in .env:

DATABASE_BACKEND=neo4j
NEO4J_URI=neo4j+s://xxx.databases.neo4j.io
NEO4J_USER=neo4j
NEO4J_PASSWORD=your_password
NEO4J_DATABASE=neo4j

See .env.mind.example for all options.

MCP Server

Claude Code

Add to .mcp.json in your project:

{
  "mcpServers": {
    "mind": {
      "command": "python3",
      "args": ["-m", "mcp.server"],
      "cwd": "/path/to/mind-mcp"
    }
  }
}

Available Tools

Tool	Description
graph_query	Semantic search across the graph
procedure_start	Start a structured dialogue
procedure_continue	Continue dialogue with answer
procedure_list	List available dialogues
doctor_check	Run health checks
agent_list	List work agents
agent_spawn	Spawn a work agent
task_list	List pending tasks

Project Structure

.runtime/
├── PRINCIPLES.md          # How to work
├── FRAMEWORK.md           # Navigation guide
├── config.yaml            # Mind config
├── database_config.yaml   # Database settings
├── agents/                # Agent postures
├── skills/                # Executable capabilities
├── procedures/            # Structured dialogues
├── state/                 # SYNC files
└── runtime/                  # Python runtime
    ├── physics/           # Graph physics
    ├── graph/             # Graph operations
    ├── connectome/        # Dialogue runner
    ├── infrastructure/    # DB adapters
    └── traversal/         # Traversal logic

Requirements

• Python 3.10+
• Neo4j or FalkorDB
• Optional: OpenAI API key (for embeddings)

License

MIT