DAAP MCP Server

A lightweight MCP (Model Context Protocol) server for CSV analysis, deployable to AWS Lambda.

Features

• CSV Analysis: Analyze CSV files with row/column counts
• S3 CSV Tools: Read and analyze CSV files directly from AWS S3
• AWS Lambda Deployment: Serverless deployment with API Gateway
• Cursor Integration: Works with Cursor IDE via MCP proxy
• Configurable: Environment-based configuration

Configuration

Environment Variables

• MCP_SERVER_URL: Your deployed MCP server URL (default: current Lambda URL)
• AWS_PROFILE: AWS profile for local development
• AWS_REGION: AWS region (default: eu-central-1)

Setup

1. Copy environment template:

   cp env.example .env

2. Update your MCP server URL in .env:

   MCP_SERVER_URL=https://your-api-gateway-id.execute-api.eu-central-1.amazonaws.com/prod/mcp
   

Centralized Configuration

All deployment scripts use centralized configuration from config.env:

• FUNCTION_NAME: Lambda function name
• API_GATEWAY_NAME: API Gateway name
• AWS_REGION: AWS region
• MCP_SERVER_URL: MCP server endpoint URL

Usage

Local Testing

# Test Lambda handler
uv run lambda_handler.py

# Test MCP proxy
uv run test_proxy.py

S3 CSV Tools

The server includes a simple tool for getting basic information from CSV files stored in AWS S3:

• analyze_s3_csv: Basic info - count, columns, and sample 50 records

📊 Ultra Simple (KISS):

• Pandas integration - uses pd.read_csv() directly with S3 URLs
• No analysis - just count, columns, and sample data
• Easy to use - just provide bucket name and file key

Example Usage:

# Get basic info from a CSV file in S3
analyze_s3_csv("my-bucket", "data/sales.csv")

Deployment

# Create deployment package
./create_deployment_package.sh

# Deploy via GitHub Actions (manual trigger)
# Or upload mcp-server-deployment.zip to AWS Lambda

Cleanup

# Local cleanup (requires confirmation)
./cleanup.sh

# GitHub Actions cleanup (manual trigger with confirmation)
# Go to Actions → Cleanup AWS Resources → Run workflow
# Type "DELETE" when prompted

Cursor Integration

The MCP proxy automatically connects to your deployed server using the configured URL.

Project Structure

├── lambda_handler.py          # AWS Lambda entry point
├── mcp_proxy.py              # MCP proxy for Cursor
├── server.py                 # MCP server configuration
├── tools/
│   ├── __init__.py           # Auto-registers all tools
│   ├── csv_tools.py          # Local CSV analysis tools
│   └── s3_csv_tools.py       # S3 CSV analysis tools
├── utils/
│   ├── file_reader.py        # Local file utilities
│   └── s3_csv_processor.py   # S3 CSV processing utilities
├── create_deployment_package.sh  # Deployment script
└── .github/workflows/deploy.yml  # GitHub Actions

fast-mcp-demo