aspenas/readai-mcp-server

Read.ai MCP Server

A production-ready MCP (Model Context Protocol) server that integrates with Read.ai's webhook system to store and query meeting data.

Features

• Webhook Integration: Receives and stores Read.ai meeting data via webhook
• SQLite Database: Persistent storage of meetings, participants, action items, topics, transcripts
• 7 MCP Tools: Natural language querying through Tasklet/Claude
• FastAPI: Modern, fast web framework with automatic API documentation
• Docker Support: Easy deployment with Docker and Docker Compose
• Railway/Fly.io Ready: One-click deployment configurations included

MCP Tools Available

1. search_meetings - Search meetings by title, participant, or date range
2. get_meeting_details - Get full details of a specific meeting
3. get_action_items - Retrieve action items with optional filtering
4. search_transcript - Search meeting transcripts by content or speaker
5. get_participant_meetings - Get all meetings for a specific participant
6. get_topics - Get topics discussed across meetings
7. get_meeting_summary - Get summary statistics and insights

Quick Start

Local Development

1. Clone the repository:

git clone https://github.com/candlefish-ai/readai-mcp-server.git
cd readai-mcp-server

2. Install dependencies:

pip install -r requirements.txt

3. Copy environment file:

cp .env.example .env

4. Run the server:

uvicorn readai_mcp.main:app --reload

The server will be available at http://localhost:8000

Docker Deployment

docker-compose up -d

Railway Deployment

1. Install Railway CLI:

npm install -g railway

2. Initialize and deploy:

railway login
railway init
railway variables set WEBHOOK_TOKEN=1df1a04b5ec6ba110262d321e8cb1202
railway up

3. Get your deployment URL:

railway domain

Read.ai Webhook Configuration

1. Log into Read.ai
2. Navigate to Settings → Integrations → Webhooks
3. Click "Add Webhook"
4. Configure:
   • URL: https://your-deployment-url/webhook
   • Token: 1df1a04b5ec6ba110262d321e8cb1202
   • Event: meeting_end

Connecting to Tasklet/Claude

Once deployed, connect via MCP:

setup_mcp_server_connection(
    displayName="Read.ai Meetings",
    serverUrl="https://your-deployment-url/mcp"
)

API Documentation

FastAPI automatically generates interactive API documentation:

• Swagger UI: http://your-server/docs
• ReDoc: http://your-server/redoc

Testing

Run the test suite:

pytest tests/ -v

Test webhook manually:

curl -X POST https://your-server/webhook \
  -H "Content-Type: application/json" \
  -H "X-Webhook-Token: 1df1a04b5ec6ba110262d321e8cb1202" \
  -d '{
    "session_id": "test-123",
    "trigger": "meeting_end",
    "title": "Test Meeting",
    "start_time": "2025-11-01T10:00:00Z",
    "end_time": "2025-11-01T11:00:00Z",
    "owner": {"name": "Test User", "email": "test@example.com"},
    "participants": [],
    "summary": "Test summary",
    "action_items": [],
    "key_questions": [],
    "topics": [],
    "report_url": "https://test.com"
  }'

Example Queries (via MCP)

Once connected to Tasklet/Claude, you can ask:

• "Search for meetings with Jesse about PromoterOS"
• "Get all action items from last week's meetings"
• "Find transcript mentions of Watsonx Orchestrate"
• "Show me all meetings Alice participated in this month"
• "What topics were discussed in product meetings?"
• "Give me a summary of meeting activity for October"

Project Structure

readai-mcp-server/
├── readai_mcp/
│   ├── __init__.py
│   ├── main.py           # FastAPI app + MCP server
│   ├── database.py       # SQLAlchemy models & DB operations
│   ├── models.py         # Pydantic models
│   ├── mcp_tools.py      # MCP tool implementations
│   └── config.py         # Configuration
├── tests/
│   ├── test_webhook.py   # Webhook endpoint tests
│   └── test_mcp_tools.py # MCP tools tests
├── Dockerfile
├── docker-compose.yml
├── railway.json          # Railway deployment config
├── fly.toml              # Fly.io deployment config
├── requirements.txt
└── README.md

Environment Variables

Monitoring

View logs (Railway):

railway logs --follow

View logs (Docker):

docker logs readai-mcp-server -f

Check database:

sqlite3 meetings.db "SELECT * FROM meetings LIMIT 5;"

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

MIT License - see LICENSE file for details

Support

For issues or questions:

• GitHub Issues: github.com/candlefish-ai/readai-mcp-server/issues
• Linear Issue: CAN-17

Acknowledgments

• Built for PromoterOS/Facture discovery sessions
• Powered by FastAPI and MCP
• Integrated with Read.ai's webhook system