🧠 OmniMind MCP

A sophisticated cognitive reasoning system that helps AI assistants think through complex problems using structured methodologies. Built on the Model Context Protocol (MCP).

✨ Features

🎯 Core Capabilities

• Multi-Agent Thinking Engine - 7 specialized thinking engines working together
• Bias Detection - Real-time detection of 27 cognitive biases
• Logic Validation - Formal and informal fallacy detection
• Assumption Tracking - Track and validate underlying assumptions
• Blind Spot Detection - Identify missing perspectives and gaps
• Long-Term Memory - Learn from past sessions and improve over time
• Pattern Recognition - Recognize productive and unproductive thinking patterns

🛡️ Shield Systems

| Shield | Description | |--------|-------------| | Bias Detector | Detects 27 cognitive biases with real pattern matching | | Logic Validator | Identifies 21+ logical fallacies | | Assumption Tracker | Tracks assumption dependencies and validation | | Blind Spot Radar | Finds missing perspectives (temporal, scale, domain, etc.) |

🧠 Thinking Engines

| Engine | Purpose | |--------|---------| | Analyzer | Problem classification and decomposition | | Explorer | UCB1-based path exploration | | Challenger | 7 challenge modes (steelman, premortem, etc.) | | Synthesizer | 5 synthesis modes (convergent, divergent, emergent) | | Calibrator | Confidence calibration | | Decider | Multi-factor decision making | | Reflector | Meta-cognitive analysis |

📊 Visualization

• Graph Renderer - ASCII/Unicode tree, mindmap, detailed views
• Progress Tracker - Phase-based progress with ETA estimation
• Quality Dashboard - 5-dimension quality scoring with health checks

🚀 Quick Start

Installation

npm install -g omnimind-mcp

Running the Server

# Start the MCP server
omnimind-mcp

# With custom database
OMNIMIND_DB_PATH=/path/to/db omnimind-mcp

Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "omnimind": {
      "command": "npx",
      "args": ["omnimind-mcp"],
      "env": {
        "OMNIMIND_DB_PATH": "./data/omnimind.db",
        "OMNIMIND_LOG_LEVEL": "info"
      }
    }
  }
}

📖 Available Tools

omnimind.think

Engage the cognitive engine to think deeply about a problem.

Parameters:

• input (string, required): The problem or question
• mode (enum): "auto" or "manual" (default: "auto")
• thinkingStyle (enum): analytical, creative, systematic, adversarial, etc.
• sessionId (string): Continue existing session
• action (enum): analyze, think, challenge, synthesize, calibrate, decide, reflect, branch

Example:

{
  "input": "How do I design a scalable architecture?",
  "mode": "manual",
  "thinkingStyle": "analytical",
  "action": "analyze"
}

omnimind.visualize

Visualize the thought graph for a session.

Parameters:

• sessionId (string, required): Session to visualize
• format (enum): ascii, unicode, compact, detailed, mindmap (default: "unicode")
• showPruned (boolean): Include pruned thoughts

omnimind.progress

Display session progress with phase completion and ETA.

Parameters:

• sessionId (string, required)
• compact (boolean): Show compact indicator only

omnimind.quality

Show quality dashboard with scores and health checks.

Parameters:

• sessionId (string, required)
• compact (boolean): Show compact indicator only

omnimind.analyze

Run comprehensive analysis (biases, logic, assumptions, blind spots).

Parameters:

• sessionId (string, required)
• analysisType (enum): all, biases, logic, assumptions, blindspots (default: "all")

omnimind.sessions

Manage thinking sessions.

Parameters:

• action (enum): list, get, delete, stats (required)
• sessionId (string): For get/delete actions
• limit (number): For list action (default: 10)

omnimind.memory

Access long-term memory and patterns.

Parameters:

• action (enum): search, recommend, patterns, stats (required)
• query (string): Search query
• domain (string): Domain filter
• problemType (string): Problem type filter

🏗️ Architecture

OmniMind MCP
├── Core Layer
│   ├── CognitiveCore (Main orchestrator)
│   ├── ThoughtGraph (Graph engine)
│   ├── AutoPilot (Decision automation)
│   └── ModeRouter (Input routing)
│
├── Engine Layer (7 Thinking Engines)
│   ├── Analyzer (Classification)
│   ├── Explorer (UCB1 exploration)
│   ├── Challenger (Devil's advocate)
│   ├── Synthesizer (Insight combination)
│   ├── Calibrator (Confidence tuning)
│   ├── Decider (Decision making)
│   └── Reflector (Meta-cognition)
│
├── Shield Layer (4 Protection Systems)
│   ├── BiasDetector (27 biases)
│   ├── LogicValidator (21 fallacies)
│   ├── AssumptionTracker (Validation)
│   └── BlindSpotRadar (Gap detection)
│
├── Memory Layer (3 Systems)
│   ├── SessionStore (SQLite persistence)
│   ├── LongTermMemory (Cross-session learning)
│   └── PatternLibrary (Pattern recognition)
│
└── Visualizer Layer (3 Renderers)
    ├── GraphRenderer (Multiple formats)
    ├── ProgressTracker (Progress bars)
    └── QualityDashboard (Metrics)

🔧 Development

Prerequisites

• Node.js 22+
• TypeScript 5.6+
• SQLite (for session persistence)

Setup

# Clone repository
git clone https://github.com/omnimind/omnimind-mcp.git
cd omnimind-mcp

# Install dependencies
npm install

# Build
npm run build

# Run tests
npm test

# Development mode
npm run dev

Project Structure

src/
├── core/           # Core orchestration
├── engines/        # 7 Thinking engines
├── shields/        # 4 Protection shields
├── memory/         # 3 Memory systems
├── visualizer/     # 3 Visualization tools
├── types/          # Type definitions
├── utils/          # Utilities
├── server.ts       # MCP server
└── index.ts        # Entry point

🧪 Testing

# Run all tests
npm test

# Run with coverage
npm run test:coverage

# Watch mode
npm run test:watch

📝 Environment Variables

| Variable | Description | Default | |----------|-------------|---------| | OMNIMIND_DB_PATH | SQLite database path | ./data/omnimind.db | | OMNIMIND_LOG_LEVEL | Logging level | info | | OMNIMIND_BACKUP_INTERVAL | Backup interval (ms) | 3600000 (1 hour) | | OMNIMIND_MAX_BACKUPS | Maximum backups to keep | 5 |

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request

📄 License

MIT License - see LICENSE file

🙏 Acknowledgments

• Built on the Model Context Protocol
• Inspired by cognitive science research on structured thinking
• UCB1 algorithm from multi-armed bandit theory

📞 Support

• GitHub Issues: github.com/omnimind/omnimind-mcp/issues
• Documentation: docs.omnimind.dev

Made with 🧠 by the OmniMind Team