vLLM MCP Server

Instructions

Implementation Reference

• src/vllm_mcp_server/tools/models.py:40-73 (handler)

  Main handler function get_model_info that validates model_id argument, calls VLLMClient to fetch model information, and formats the response as TextContent with JSON-formatted model details.

  async def get_model_info(arguments: dict[str, Any]) -> list[TextContent]:
      """
      Get detailed information about a specific model.
  
      Args:
          arguments: Dictionary containing:
              - model_id: The ID of the model to get info for
  
      Returns:
          List of TextContent with detailed model information.
      """
      model_id = arguments.get("model_id")
      if not model_id:
          return [TextContent(type="text", text="Error: No model_id provided")]
  
      try:
          async with VLLMClient() as client:
              model_info = await client.get_model_info(model_id)
  
              if not model_info:
                  return [
                      TextContent(type="text", text=f"Model '{model_id}' not found on the server.")
                  ]
  
              # Format model info
              result = f"## Model: {model_id}\n\n"
              result += "```json\n"
              result += json.dumps(model_info, indent=2)
              result += "\n```"
  
              return [TextContent(type="text", text=result)]
  
      except VLLMClientError as e:
          return [TextContent(type="text", text=f"Error getting model info: {str(e)}")]

• src/vllm_mcp_server/utils/vllm_client.py:104-110 (helper)

  VLLMClient.get_model_info helper method that retrieves model information by listing all models and finding the one matching the given model_id.

  async def get_model_info(self, model_id: str) -> Optional[dict[str, Any]]:
      """Get information about a specific model."""
      models = await self.list_models()
      for model in models:
          if model.get("id") == model_id:
              return model
      return None

• src/vllm_mcp_server/server.py:139-151 (registration)

  Tool registration defining get_model_info with its name, description, and inputSchema that requires a model_id string parameter.

  Tool(
      name="get_model_info",
      description="Get detailed information about a specific model",
      inputSchema={
          "type": "object",
          "properties": {
              "model_id": {
                  "type": "string",
                  "description": "The ID of the model to get info for",
              },
          },
          "required": ["model_id"],
      },

• src/vllm_mcp_server/server.py:348-349 (registration)

  Handler mapping in call_tool function that routes 'get_model_info' tool calls to the get_model_info handler function.

  elif name == "get_model_info":
      return await get_model_info(arguments)