Brain Assistant MCP Server

A natural language interface for the Brain knowledge management system. Talk to your Brain system like you would talk to a helpful assistant.

🧠 What is Brain Assistant?

Brain Assistant provides a conversational interface to all Brain system tools, allowing you to:

• Use natural language instead of JSON commands
• Execute complex workflows with simple requests
• Get intelligent suggestions for tools to use
• Batch operations with preview and approval
• Maintain conversation context across interactions

🚀 Features

Natural Language Commands

Instead of:

brain_remember { "key": "meeting_notes_2024", "value": "..." }

Just say:

"Remember that we discussed API redesign in today's meeting"

Intelligent Workflows

"Create a summary of this week's work and email it to me"

The assistant will:

1. Search relevant memories and notes
2. Generate a summary
3. Create a formatted email
4. Send it (with your approval)

Smart Suggestions

"I need to organize my project notes"

Get suggestions for:

• Which tools to use
• Example commands
• Best practices

Batch Operations

"Archive all memories older than 6 months except the important ones"

• Preview changes before executing
• Intelligent filtering
• Safety checks

📦 Installation

# Clone the repository
git clone https://github.com/[your-username]/mcp-brain-assistant.git
cd mcp-brain-assistant

# Install dependencies
npm install

# Build the project
npm run build

🔧 Configuration

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "brain-assistant": {
      "command": "node",
      "args": ["/path/to/mcp-brain-assistant/dist/index.js"],
      "env": {
        "ANTHROPIC_API_KEY": "your-api-key-here"
      }
    }
  }
}

🎯 Usage

Basic Chat

brain_chat {
  "message": "What did I work on yesterday?",
  "context": {
    "currentProject": "api-redesign"
  }
}

Execute Workflows

brain_workflow {
  "workflow": "Prepare weekly status report with tasks completed and blockers",
  "requireApproval": true
}

Get Tool Suggestions

brain_suggest {
  "task": "I want to track my daily habits",
  "includeExamples": true
}

Batch Operations

brain_batch {
  "operation": "Tag all notes about machine learning with #ml tag",
  "preview": true
}

🖥️ Streamlit UI

Run the interactive UI:

npm run streamlit

Features:

• Chat interface with message history
• Visual approval system for operations
• Quick action buttons
• Tool reference panel
• Project context management

💡 Example Conversations

Morning Routine

You: "Good morning! What's on my plate today?"
Assistant: *Checks tasks, calendar, and recent work*
"Good morning! Here's your day:
- 3 high-priority tasks in the API project
- Meeting notes from yesterday need review
- 2 pending emails about the design spec
Would you like me to create today's daily note?"

Project Context Switch

You: "Switch to the mobile app project"
Assistant: *Loads project context*
"Switched to mobile app project. Last worked on: 3 days ago
- Status: In development
- Next milestone: Authentication implementation
- 5 open tasks
Would you like to see the task list?"

Smart Search

You: "Find everything about the authentication bug"
Assistant: *Searches across memories, notes, and emails*
"Found 12 items about the authentication bug:
- 3 memories with debugging notes
- 2 Obsidian notes with solutions
- 5 emails discussing the issue
- 2 tasks related to the fix
The main issue was token expiration. Solution implemented on July 20."

🤝 Human Approval System

The assistant requires approval for:

• Sending emails
• Deleting content
• Executing code
• Batch modifications

Auto-approval available for:

• Read operations
• Creating drafts
• Non-destructive searches
• Status checks

🧩 Integration with Brain Tools

The assistant seamlessly integrates with:

• brain - Core memory and execution
• brain-manager - Project management
• filesystem - File operations
• todo-manager - Task tracking
• obsidian - Note management
• Claude's built-in tools (web search, gmail, etc.)

🔐 Security

• All operations respect Brain system permissions
• Approval required for sensitive operations
• Conversation history stored locally
• No data sent to external services (except Anthropic for NLP)

🚧 Roadmap

• [ ] Voice input support
• [ ] Scheduled workflows
• [ ] Learning from user patterns
• [ ] Multi-user support
• [ ] Mobile app
• [ ] Webhook integrations

🤔 Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines.

📄 License

MIT License - see LICENSE for details.

---

Transform your Brain system from a powerful tool into an intelligent companion! 🧠✨