Brain MCP

🚀 Brain-MCP

Memory Control Protocol - An AI memory and thinking management system that simulates the human-like thinking and memory process.

Brain-MCP offers a comprehensive memory management system for AI, supporting short-term memory, long-term memory, and thinking process management.

❌ Without Brain-MCP

AI lacks the ability to retain memories persistently. Each conversation starts afresh, leading to the following issues:

❌ Inability to recall previous conversation content. ❌ Inability to establish associations between knowledge. ❌ Inability to conduct continuous thinking processes. ❌ Inability to learn and accumulate experience.

✅ Using Brain-MCP

Brain-MCP provides AI with a complete memory management system, supporting short-term memory, long-term memory, and thinking process management:

✅ Short-term Memory: A FIFO queue-based temporary cache with configurable capacity. ✅ Long-term Memory: A graph-based persistent storage system supporting associations and searches. ✅ Thinking Process Management: A complete thinking chain management system supporting multiple types of thinking nodes. ✅ Automatic Persistence: Every memory modification is automatically saved. ✅ Cognitive Modes: Supports multiple modes such as analysis, intuition, creation, criticism, and metacognition.

🚀 Quick Start

Requirements

• Node.js >= v18.0.0
• Claude Desktop, Cursor, VSCode, or other MCP clients

Configuration

{
  "mcpServers": {
    "brain": {
      "command": "npx",
      "args": ["-y", "digital-brain-mcp"],
      "env": {
        "BRAIN_MCP_STORAGE_PATH":"./memory_data"
      }
    }
  }
}

🔨 Available Tools

Brain-MCP provides the following tools for AI:

Short-term Memory Tools

• addShortTermMemory: Add temporary memory to the short-term cache.
• getShortTermMemory: Retrieve all short-term memories (newest first).
• clearShortTermMemory: Clear all short-term memories.

Long-term Memory Tools

• addLongTermMemory: Create a new persistent memory node.
• getLongTermMemory: Retrieve memory and its associations by ID.
• searchLongTermMemory: Search for memories by keyword.
• updateLongTermMemory: Update memory content and associations.
• deleteLongTermMemory: Delete a memory node.
• getAssociations: Retrieve direct associations of a node.

Thinking Process Tools

• startThoughtProcess: Initiate a new thinking chain.
• addThought: Add a thinking node to the thinking chain.
• branchThought: Create a branch from an existing thought.
• evaluateThought: Evaluate and update the confidence level of a thought.
• completeThoughtProcess: Complete the thinking process and store the conclusion.
• getCurrentThoughtChain: Retrieve the current state of the thinking chain.
• pauseThinking: Pause an active thinking process.
• resumeThinking: Resume a paused thinking process.
• switchCognitiveMode: Switch cognitive modes.
• getOptimalModeForTask: Get the recommended optimal cognitive mode for a task.
• getThinkingProgress: Retrieve the progress statistics of the thinking chain.
• getActiveChains: Retrieve all active thinking chains.
• getThinkingStats: Retrieve overall statistics of the thinking process.

System Management Tools

• saveMemory: Manually save memories to disk.
• getMemoryStats: Retrieve memory system statistics.

💻 Usage Examples

Basic Memory Operations

// Add short-term memory
await mcp.addShortTermMemory("User inquired about the weather.");

// Add long-term memory and establish associations
const memoryId = await mcp.addLongTermMemory("It's hot in Beijing in summer", ["weather-memory-id"]);

// Search for memories
const results = await mcp.searchLongTermMemory("hot");

Thinking Process Management

// Start a thinking process
const chainId = await mcp.startThoughtProcess("Design a user-friendly product interface");

// Add thinking nodes
await mcp.addThought(chainId, "Think from the user's perspective about the main usage scenarios", "analysis");
await mcp.addThought(chainId, "Adopt a minimalist design style", "decision");

// Create a branch to explore alternative solutions
const branchId = await mcp.branchThought(thoughtId, "Try an innovative interaction mode");

// Complete the thinking process
await mcp.completeThoughtProcess(chainId, "Choose a balanced solution: minimalist design with moderate innovation");

🛟 Tips

Automatic Saving

All memory modification operations are automatically saved to local storage to ensure data integrity.

Memory Association

Use the association feature to establish connections between knowledge, enabling deep retrieval and intelligent recommendation.

Cognitive Modes

Choose the appropriate cognitive mode based on the task type:

• Analysis Mode: In-depth reasoning and multi-angle analysis.
• Intuition Mode: Quick association and pattern recognition.
• Creation Mode: Divergent thinking and concept combination.
• Criticism Mode: Questioning, verification, and logical checking.
• Metacognition Mode: Thinking about one's own thinking process.

💻 Development

Clone the project and install dependencies:

git clone https://github.com/DDguan2010/brain-mcp.git
cd brain-mcp
npm install

Build the project:

npm run build

Run the server:

npm start

CLI Parameters

brain-mcp accepts the following CLI parameters:

• --port : Server port (default: 3000).
• --storage-path : Storage path (default: ./memory_data).
• --auto-save-interval : Automatic save interval (default: 5 minutes).

Example:

npm start -- --port 8080 --storage-path ./my_memory

Environment Variables

The following environment variables can be used:

• NODE_ENV: Runtime environment (development/production).
• BRAIN_MCP_STORAGE_PATH: Storage path.
• BRAIN_MCP_AUTO_SAVE_INTERVAL: Automatic save interval.

Example:

NODE_ENV=production BRAIN_MCP_STORAGE_PATH=./data npm start

🏗️ Architecture Design

brain-mcp/
├── src/
│   ├── types.ts              # Type definitions
│   ├── config.ts             # Configuration management
│   ├── short-term-memory.ts  # Short-term memory management
│   ├── long-term-memory.ts   # Long-term memory management
│   ├── thinking-process.ts   # Thinking process management
│   ├── storage.ts            # File persistence
│   └── brain-mcp.ts          # Main module
└── server.ts                 # MCP server

📄 License

MIT License

🤝 Contribution

Feel free to submit issues and pull requests!

Made by 0.2Studio