human-mcp

Instructions

Implementation Reference

• human_mcp/mcp_server.py:83-101 (handler)

  The handler function for human_mouth_tool. It generates a unique task ID, formats the utterance as a task instruction, adds it to the database using db_utils, polls asynchronously for completion using wait_for_task_completion, and returns the human-provided response.

  @mcp.tool()
  async def human_mouth_tool(utterance: str, ctx: Context) -> Dict[str, str]:
      """人間が口を使って指定された言葉を発話します。"""
      task_id = str(uuid.uuid4())
      formatted_utterance = f"👄 口を使って発話: {utterance}"
  
      # タスクをデータベースに追加
      db_utils.add_task(task_id, formatted_utterance)
  
      # ログ出力
      sys.stderr.write(f"Human task created: {task_id}. Waiting for completion...\n")
  
      # 結果を待機（非同期ポーリング）
      result = await wait_for_task_completion(task_id)
  
      # ログ出力
      sys.stderr.write(f"Human task {task_id} completed.\n")
  
      return {"response": result}

• human_mcp/tools.py:41-58 (schema)

  JSON schema definition for the human_mouth_tool, specifying input 'utterance' parameter and output 'response' field.

  {
      "name": "human_mouth_tool",
      "description": "人間が口を使って指定された言葉を発話します。",
      "input_schema": {
          "type": "object",
          "properties": {
              "utterance": {"type": "string", "description": "発話する内容"}
          },
          "required": ["utterance"]
      },
      "output_schema": {
          "type": "object",
          "properties": {
              "response": {"type": "string", "description": "発話に対する応答"}
          },
          "required": ["response"]
      }
  },

• human_mcp/mcp_server.py:83-83 (registration)

  The @mcp.tool() decorator registers the human_mouth_tool with the FastMCP server.

  @mcp.tool()