Pensieve MCP Server

Instructions

Implementation Reference

• mcp_server/server.py:30-48 (handler)

  The core logic implementation of the save_conversation tool which saves the conversation to a JSON file and cache.

  def save_conversation(conversation_id: str, messages: List[Dict[str, Any]], metadata: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
      """대화를 파일 시스템에 저장"""
      conversation_data = {
          "id": conversation_id,
          "messages": messages,
          "metadata": metadata or {},
          "created_at": datetime.now().isoformat(),
          "updated_at": datetime.now().isoformat()
      }
      
      # 파일로 저장
      file_path = STORAGE_DIR / f"{conversation_id}.json"
      with open(file_path, 'w', encoding='utf-8') as f:
          json.dump(conversation_data, f, ensure_ascii=False, indent=2)
      
      # 캐시에도 저장
      conversation_cache[conversation_id] = conversation_data
      
      return conversation_data

• mcp_server/server.py:139-155 (registration)

  MCP tool registration for save_conversation in the list_tools method.

  Tool(
      name="save_conversation",
      description="대화 내역을 저장합니다",
      inputSchema={
          "type": "object",
          "properties": {
              "conversation_id": {
                  "type": "string",
                  "description": "대화 ID (없으면 자동 생성)"
              },
              "messages": {
                  "type": "array",
                  "description": "저장할 메시지 목록",
                  "items": {
                      "type": "object",
                      "properties": {
                          "role": {"type": "string", "enum": ["user", "assistant", "system"]},