MCP Server Template

""" Example tool demonstrating DataServiceClient usage This tool shows how to use the injected DataServiceClient from oxsci-oma-mcp to call data service endpoints with minimal boilerplate code. Note: IDataServiceClient and require_data_service are provided by oxsci-oma-mcp >= 0.2.0 """ from fastapi import Depends from pydantic import BaseModel, Field from oxsci_oma_mcp import ( oma_tool, require_context, IMCPToolContext, IDataServiceClient, require_data_service, ) class ExampleDataServiceRequest(BaseModel): """Request model for example_data_service_tool""" overview_id: str = Field( ..., description="Overview ID to fetch sections from data service" ) user_id: str = Field(default=None, description="Optional user ID for filtering") class ExampleDataServiceResponse(BaseModel): """Response model for example_data_service_tool""" overview_id: str = Field(..., description="Overview ID that was fetched") sections: list = Field(..., description="List of sections from data service") metadata: dict = Field( default_factory=dict, description="Additional metadata about the operation" ) @oma_tool( description="Example tool demonstrating DataServiceClient usage with data service integration", version="1.0.0", enable=False, # Set to True to enable this tool ) async def example_data_service_tool( request: ExampleDataServiceRequest, context: IMCPToolContext = Depends(require_context), data_service: IDataServiceClient = Depends(require_data_service), ) -> ExampleDataServiceResponse: """ Example tool demonstrating simplified data service calls using DataServiceClient. This tool shows: - How to inject IDataServiceClient via Depends(require_data_service) - How to make clean API calls without boilerplate code - How authentication is automatically forwarded - How to handle responses from data service Comparison with old pattern: OLD: service_client = ServiceClient(f"{config.SERVICE_NAME}-get_sections") sections = await service_client.call_service( target_service_url=config.DATA_SERVICE_URL, method="GET", endpoint=f"/article_structured_contents/overviews/{overview_id}/sections", timeout=30, query_params={**({"user_id": user_id} if user_id else {})}, ) NEW: sections = await data_service.call( method="GET", endpoint=f"/article_structured_contents/overviews/{overview_id}/sections", query_params={"user_id": user_id} if user_id else {}, ) Args: request: Tool request parameters context: MCP context for accessing shared data (auto-injected by tool_router) data_service: Data service client (auto-injected by tool_router) Returns: ExampleDataServiceResponse with sections data and metadata Example: POST /tools/example_data_service_tool { "arguments": { "overview_id": "overview_123", "user_id": "user_456" }, "context": {} } """ # Get user_id from request or context user_id = request.user_id or context.get_shared_data("user_id", None) # ==================== Clean and Simple Data Service Call ==================== # No need to: # - Create ServiceClient # - Pass target_service_url # - Pass user_request for authentication # All handled automatically by DataServiceClient! try: sections = await data_service.call( method="GET", endpoint=f"/article_structured_contents/overviews/{request.overview_id}/sections", query_params={"user_id": user_id} if user_id else {}, timeout=30, ) except Exception as e: # Handle error (endpoint might not exist in your data service) sections = [] metadata = { "error": str(e), "note": "This is a demonstration tool. Adjust the endpoint to match your data service API.", } else: metadata = { "overview_id": request.overview_id, "user_id": user_id, "section_count": len(sections) if isinstance(sections, list) else 0, } # ==================== Store in context for tool chaining ==================== context.set_shared_data("last_overview_id", request.overview_id) context.set_shared_data("last_sections", sections) # ==================== Return response ==================== return ExampleDataServiceResponse( overview_id=request.overview_id, sections=sections, metadata=metadata ) # ==================== Benefits Summary ==================== """ Benefits of using IDataServiceClient (from oxsci-oma-mcp): ✅ **Automatic Injection**: No manual client creation ✅ **Pre-configured**: DATA_SERVICE_URL auto-loaded from config ✅ **Auth Forwarding**: Authentication automatically passed through ✅ **Single Instance**: Reused across multiple calls in same request ✅ **Clean API**: Only specify method and endpoint ✅ **Less Code**: Reduces boilerplate by ~60% ✅ **Consistent Pattern**: Same as require_context from oxsci-oma-mcp Code Reduction Example: Before: 6 lines to make a data service call After: 3 lines to make the same call Reduction: 50% less code, 100% clearer intent """