Expanding Managed Agents in Gemini API: background tasks, remote MCP and more
Google expands Managed Agents in the Gemini Interactions API: long-running background execution (start with background:true, poll/stream/reconnect by ID), direct connection to remote MCP servers alongside built-in sandbox tools, custom function calling, and credential refreshing across interactions.
Marked important, but published more than 30 days ago, so it moved down to this band.
On this page11 sections
The editors' notes below — why it matters, the context, the learning path — are written in Chinese.
Record
The signal
#Google announced new capabilities for Managed Agents in the Gemini Interactions API. With managed agents, a single endpoint call has Gemini handle reasoning, code execution, package installation, file management and web information inside an isolated cloud sandbox. New in this update: long-running background execution (pass background:true to run asynchronously on the server; the API returns an ID for polling, streaming progress, or reconnecting later), remote MCP server integration (connect managed agents directly to remote Model Context Protocol servers, mixing remote tools with built-in capabilities like Google Search and code execution), custom function calling, and credential refreshing across interactions.
Why now
Why it matters
#托管式长时程智能体正在成为大厂平台的标配：执行沙箱、后台任务、工具接入都向平台侧收敛，MCP 又拿下一个主流平台的官方支持。团队需要重新评估「自建智能体 runtime」还是「用平台托管」——这条边界正在快速移动。
Background
Technical context
#后台执行解决的是长时程任务的连接脆弱性：同步 HTTP 模式下任务时长受连接寿命限制，异步模式把状态管理移到服务端（凭 ID 轮询/重连），这是长时程智能体状态管理的平台化实现。远程 MCP 直连则把「模型 ↔ 私有工具」的集成从自建代理中间件简化为协议标准件——与内置工具（搜索、代码执行）在同一沙箱内混用。
Audience
Who should care
#Next
Learning path
#- 先读「工具使用与函数调用」与 MCP 的基本概念，理解托管接入与自建代理的差异
- 对照 ChatGPT Work 的产品形态，理解长时程后台执行为什么是标配能力
- 评估自己产品里哪些工具接入可以从自建中间件迁移到 MCP 标准件
Network
Where it sits in the network
#How this signal connects to neighbouring technologies, skills and background knowledge. Follow any link to keep exploring.
Further thought
Follow-up questions
#- 远程 MCP 接入的鉴权与权限边界如何配置？沙箱内的爆炸半径怎么控制？
- 后台任务的时长上限、计费模型与并发配额是什么？
- 与自建智能体 runtime 相比，托管形态在可观测性上让渡了什么？
Publication record
Featured in
#The briefings that carried this signal, newest first; the last row is the weekly review of its week. The tag is the section it ran in.
技术简报 - 2026-08-01
Worth tracking技术简报 - 2026-07-28
Immediate attentionWeekly review · Jul 6 – Jul 12
Worth tracking
Sources