vLLM MCP Server

Instructions

Implementation Reference

• src/vllm_mcp_server/tools/server_control.py:616-660 (handler)

  Main implementation of get_vllm_logs function that fetches container logs using docker/podman runtime. Takes optional container_name and tail parameters, checks if container exists, runs the logs command, and returns formatted output with tips for following logs in real-time.

  async def get_vllm_logs(arguments: dict[str, Any]) -> list[TextContent]:
      """
      Get logs from a vLLM container.
  
      Args:
          arguments: Dictionary containing:
              - container_name: Name of container (default: from settings)
              - tail: Number of lines to show (default: 50)
              - follow: Whether to show note about following (default: False)
  
      Returns:
          List of TextContent with container logs.
      """
      settings = get_settings()
      
      platform_info = await get_platform_info()
      if not platform_info.runtime_running:
          runtime_name = platform_info.container_runtime.value.capitalize() if platform_info.container_runtime != ContainerRuntime.NONE else "Container runtime"
          return [TextContent(type="text", text=f"❌ Error: {runtime_name} is not running.")]
  
      runtime_cmd = _get_runtime_cmd(platform_info.container_runtime)
      container_name = arguments.get("container_name", settings.container_name)
      tail = arguments.get("tail", 50)
  
      if not await _is_container_exists(container_name, platform_info.container_runtime):
          return [TextContent(
              type="text",
              text=f"❌ Container '{container_name}' does not exist."
          )]
  
      exit_code, stdout, stderr = await _run_command(
          [runtime_cmd, "logs", "--tail", str(tail), container_name]
      )
      
      if exit_code != 0:
          return [TextContent(type="text", text=f"❌ Error getting logs: {stderr}")]
      
      # Combine stdout and stderr (vLLM logs to stderr)
      logs = stdout + stderr
      
      return [TextContent(
          type="text",
          text=f"## Logs for '{container_name}' (last {tail} lines)\n\n```\n{logs}\n```\n\n"
               f"💡 **Tip:** Run `{runtime_cmd} logs -f {container_name}` in terminal to follow logs in real-time."
      )]

• src/vllm_mcp_server/server.py:267-283 (registration)

  Tool registration defining the get_vllm_logs tool schema with name, description, and inputSchema properties (container_name as string, tail as integer with default 50).

      name="get_vllm_logs",
      description="Get logs from a vLLM container to check loading progress or errors",
      inputSchema={
          "type": "object",
          "properties": {
              "container_name": {
                  "type": "string",
                  "description": "Name of the container",
              },
              "tail": {
                  "type": "integer",
                  "description": "Number of log lines to show",
                  "default": 50,
              },
          },
      },
  ),

• src/vllm_mcp_server/server.py:361-362 (registration)

  Handler invocation that calls get_vllm_logs(arguments) when the tool is invoked via the MCP server.

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

• src/vllm_mcp_server/server.py:26-33 (registration)

  Import statement that brings get_vllm_logs from vllm_mcp_server.tools.server_control module.

  from vllm_mcp_server.tools.server_control import (
      get_platform_status,
      get_vllm_logs,
      list_vllm_containers,
      restart_vllm,
      start_vllm,
      stop_vllm,
  )

• src/vllm_mcp_server/tools/__init__.py:5-29 (registration)

  Module export of get_vllm_logs function, making it available for import by other parts of the codebase.

  from vllm_mcp_server.tools.server_control import (
      get_platform_info,
      get_platform_status,
      get_vllm_logs,
      list_vllm_containers,
      restart_vllm,
      start_vllm,
      stop_vllm,
  )
  from vllm_mcp_server.tools.benchmark import run_benchmark
  
  __all__ = [
      "handle_chat",
      "handle_complete",
      "list_models",
      "get_model_info",
      "start_vllm",
      "stop_vllm",
      "restart_vllm",
      "list_vllm_containers",
      "get_vllm_logs",
      "get_platform_info",
      "get_platform_status",
      "run_benchmark",
  ]