This example demonstrates how to use EnrichMCP with a real SQLite database, showcasing:

• Lifespan management: Database connection lifecycle
• Context injection: Accessing database from resolvers and resources
• Progress reporting: Visual feedback during operations
• Logging: Structured logging throughout the API

• SQLite database with automatic schema creation
• Sample data initialization
• Context injection in all resolvers and resources
• Proper error handling and logging
• Progress reporting for long operations

python app.py

This will:

1. Create a SQLite database (shop.db) in the current directory
2. Initialize the schema with tables for users, products, orders
3. Insert sample data if the database is empty
4. Start the MCP server

The lifespan async context manager handles database setup and teardown:

@asynccontextmanager
async def lifespan(app: EnrichMCP) -> AsyncIterator[dict[str, Any]]:
    # Setup: Connect to database
    db = Database("shop.db")
    await db.connect()

    # Yield context available in handlers
    yield {"db": db}

    # Cleanup: Close connection
    await db.close()

Access the database through context injection in your resources or resolvers:

from fastmcp import Context

@app.retrieve
async def get_user(user_id: int, ctx: Context) -> User:
    # Access database from lifespan context
    db = ctx.request_context.lifespan_context["db"]

    # Use logging
    await ctx.info(f"Fetching user {user_id}")

    # Query database
    user_row = await db.get_user(user_id)
    return User(**user_row)

1. Real Database: All data comes from SQLite, not hardcoded lists
2. Async Operations: Database operations are async-ready
3. Context Usage: Shows logging, progress, and database access
4. Error Handling: Proper handling of missing data

• users: Customer accounts with risk scores
• products: Items for sale with fraud risk levels
• orders: Purchase transactions with status tracking
• order_products: Many-to-many relationship between orders and products