📄 中文摘要
提出了一种综合方法，用于在 Amazon Bedrock AgentCore 上构建长期运行的 MCP 服务器。首先，采用上下文消息策略，以保持服务器与客户端在延长操作期间的持续通信。其次，开发了异步任务管理框架，使 AI 代理能够启动长期进程而不阻塞其他操作。最后，展示了如何将这些策略与 Amazon Bedrock AgentCore 和 Strands Agents 结合，构建能够可靠处理复杂、耗时操作的生产级 AI 代理。
在 Amazon Bedrock AgentCore 上构建长期运行的 MCP 服务器与 Strands Agents 集成
📄 English Summary
Build long-running MCP servers on Amazon Bedrock AgentCore with Strands Agents integration
A comprehensive approach is presented for building long-running MCP servers on Amazon Bedrock AgentCore. A context message strategy is introduced to maintain continuous communication between servers and clients during extended operations. An asynchronous task management framework is developed, allowing AI agents to initiate long-running processes without blocking other operations. Finally, the integration of these strategies with Amazon Bedrock AgentCore and Strands Agents is demonstrated, resulting in production-ready AI agents capable of reliably handling complex, time-intensive operations.