MCP Complete Implementation Guide

Instructions

Implementation Reference

• server.py:248-267 (handler)

  Primary MCP handler for execute_command tool. Executes shell commands using subprocess.run, captures stdout/stderr, formats output, and returns as TextContent. Includes timeout and cwd support.

  async def execute_command(self, command: str, cwd: Optional[str] = None) -> list[types.TextContent]:
      """Execute system command"""
      try:
          result = subprocess.run(
              command,
              shell=True,
              capture_output=True,
              text=True,
              cwd=cwd or Path.cwd(),
              timeout=30,
          )
          
          output = f"Command: {command}\nOutput:\n{result.stdout}"
          if result.stderr:
              output += f"\nErrors:\n{result.stderr}"
          
          return [types.TextContent(type="text", text=output)]
      except Exception as error:
          raise Exception(f"Command execution failed: {str(error)}")

• server.js:293-311 (handler)

  Primary MCP handler for execute_command tool in JavaScript server. Uses child_process.exec to run commands, captures output, formats response, supports cwd.

  async executeCommand(command, cwd = process.cwd()) {
    const { exec } = require('child_process');
    const { promisify } = require('util');
    const execAsync = promisify(exec);
  
    try {
      const { stdout, stderr } = await execAsync(command, { cwd });
      return {
        content: [
          {
            type: 'text',
            text: `Command: ${command}\nOutput:\n${stdout}${stderr ? `\nErrors:\n${stderr}` : ''}`,
          },
        ],
      };
    } catch (error) {
      throw new Error(`Command execution failed: ${error.message}`);
    }
  }

• server.py:126-143 (registration)

  Registration of execute_command tool in Python MCP server's list_tools handler, including input schema definition.

  types.Tool(
      name="execute_command",
      description="Execute a system command (use with caution)",
      inputSchema={
          "type": "object",
          "properties": {
              "command": {
                  "type": "string",
                  "description": "The command to execute",
              },
              "cwd": {
                  "type": "string",
                  "description": "Working directory for the command",
              },
          },
          "required": ["command"],
      },
  ),

• server.js:99-115 (registration)

  Registration of execute_command tool in JavaScript MCP server's list_tools response, including input schema.

    name: 'execute_command',
    description: 'Execute a system command (use with caution)',
    inputSchema: {
      type: 'object',
      properties: {
        command: {
          type: 'string',
          description: 'The command to execute',
        },
        cwd: {
          type: 'string',
          description: 'Working directory for the command',
        },
      },
      required: ['command'],
    },
  },

• chatgpt-proxy.js:252-255 (handler)

  Handler implementation in ChatGPT proxy server for execute_command, used when bridging OpenAI tools to local execution.

  case 'execute_command':
    const { stdout, stderr } = await execAsync(args.command, { cwd: args.cwd || process.cwd() });
    return { success: true, stdout, stderr, command: args.command };