Podman MCP Server

Instructions

Implementation Reference

• main_b.py:562-567 (handler)

  Main handler implementation for exec_container tool. Parses args, runs 'podman exec <container> <command>' using run_podman helper, returns stdout or error message.

  async def exec_container(self, args: Dict[str, Any]) -> Dict[str, Any]:
      container = args.get("container")
      command = args.get("command", [])
      cmd_args = ["exec", container] + command
      result = run_podman(cmd_args)
      return {"output": result["stdout"] if result["success"] else f"Error: {result['stderr']}"}

• main_b.py:323-341 (schema)

  Input schema and registration in the tools list for the exec_container tool.

  Tool(
      name="exec_container",
      description="Execute a command inside a container",
      inputSchema={
          "type": "object",
          "properties": {
              "container": {
                  "type": "string",
                  "description": "Container name or ID"
              },
              "command": {
                  "type": "array",
                  "items": {"type": "string"},
                  "description": "Command to execute"
              }
          },
          "required": ["container", "command"]
      }
  ),

• main_b.py:459-472 (registration)

  Mapping of tool name 'exec_container' to its handler method in the dispatch dictionary used by handle_tools_call.

  tool_handlers = {
      "list_containers": self.list_containers,
      "container_info": self.container_info,
      "start_container": self.start_container,
      "stop_container": self.stop_container,
      "restart_container": self.restart_container,
      "container_logs": self.container_logs,
      "run_container": self.run_container,
      "remove_container": self.remove_container,
      "exec_container": self.exec_container,
      "list_images": self.list_images,
      "pull_image": self.pull_image,
      "container_stats": self.container_stats,
  }

• main.py:239-246 (handler)

  Alternative handler implementation using FastMCP decorator @mcp.tool, which also defines input schema via Pydantic Field. Executes podman exec similarly.

  @mcp.tool(title="Exec in container", description="Execute a command inside a container.")
  def exec_container(
      container: str = Field(...),
      command: List[str] = Field(..., description="Command to execute"),
  ) -> str:
      args = ["exec", container] + command
      result = run_podman(args)
      return result["stdout"] if result["success"] else f"Error: {result['stderr']}"

• main.py:131-150 (helper)

  Shared helper function used by exec_container (and other tools) to run podman subprocess commands and parse results.

  def run_podman(args: List[str]) -> Dict[str, Any]:
      """Run a podman command and capture output"""
      try:
          cmd = ["podman"] + args
          result = subprocess.run(
              cmd,
              capture_output=True,
              text=True,
              timeout=30
          )
          return {
              "success": result.returncode == 0,
              "stdout": result.stdout.strip(),
              "stderr": result.stderr.strip(),
              "returncode": result.returncode,
          }
      except subprocess.TimeoutExpired:
          return {"success": False, "stdout": "", "stderr": "Command timed out", "returncode": -1}
      except Exception as e:
          return {"success": False, "stdout": "", "stderr": str(e), "returncode": -1}