MCP Server Replicate

Instructions

    Args:
        model_id: Model identifier in format owner/name

    Returns:
        Detailed model information
    

Implementation Reference

• src/mcp_server_replicate/tools/model_tools.py:61-64 (registration)

  Registration of the 'get_model_details' tool using the @mcp.tool decorator, specifying the name and description.

  @mcp.tool(
      name="get_model_details",
      description="Get detailed information about a specific model.",
  )

• src/mcp_server_replicate/tools/model_tools.py:65-93 (handler)

  The handler function implements the logic to retrieve detailed model information by parsing the model_id, querying the owner's models via ReplicateClient, falling back to search if not found, and returning a Model object or raising ValueError.

  async def get_model_details(model_id: str) -> Model:
      """Get detailed information about a specific model.
      
      Args:
          model_id: Model identifier in format 'owner/model'
          
      Returns:
          Model object containing detailed model information
          
      Raises:
          RuntimeError: If the Replicate client fails to initialize
          ValueError: If the model is not found
          Exception: If the API request fails
      """
      owner, name = model_id.split("/")
      async with ReplicateClient() as client:
          # First try to find the model in the owner's models
          result = client.list_models(owner=owner)
          for model in result["models"]:
              if model["name"] == name:
                  return Model(**model)
          
          # If not found, try searching for it
          search_result = await client.search_models(model_id)
          for model in search_result["models"]:
              if f"{model['owner']}/{model['name']}" == model_id:
                  return Model(**model)
          
          raise ValueError(f"Model not found: {model_id}")