[{"data":1,"prerenderedAt":27},["ShallowReactive",2],{"nr-en-google-gemini-managed-agents-background-tasks":3},{"slug":4,"title":5,"dek":6,"date":7,"time":8,"publishedAt":9,"updated":10,"updatedAt":10,"dateFmt":11,"updatedFmt":10,"kind":12,"tier":13,"author":14,"authorName":15,"topics":16,"tracker":10,"trackerLabel":10,"headlineStat":10,"image":22,"ogImage":23,"imageAlt":5,"csv":10,"minutes":24,"words":25,"html":26},"google-gemini-managed-agents-background-tasks","Google Expands Gemini Agents with Background Execution and Remote Protocols","The Gemini API gains new capabilities for production-ready AI agents: background tasks, remote MCP integration, and credential refresh. Developers can now build reliable, asynchronous agents in an isolated sandbox.","2026-07-08","07:32","2026-07-08T07:32:00+02:00","","July 8, 2026","news","standard","ideal-syka","Ideal Syka",[17,18,19,20,21],"Gemini API","AI Agents","Managed Agents","Background Tasks","MCP","\u002Fnewsroom\u002Fimg\u002Fgoogle-gemini-managed-agents-background-tasks.webp","\u002Fog-nr\u002Fgoogle-gemini-managed-agents-background-tasks.en.png",2,414,"\u003Cp>Google has announced new features for \u003Cstrong>Managed Agents in the Gemini API\u003C\u002Fstrong>, enabling developers to build reliable, production-ready AI agents. The expansion addresses developer feedback and includes background execution, remote MCP server integration, custom functions, and credential refresh – all operating within an isolated cloud sandbox environment.\u003C\u002Fp>\n\u003Ch2>Key Takeaways\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Cstrong>Background Execution\u003C\u002Fstrong>: Agents run asynchronously on the server; the API returns an ID immediately instead of keeping HTTP connections open\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Remote MCP Servers\u003C\u002Fstrong>: Direct connection to private databases and internal APIs without custom proxy middleware\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Custom Functions\u003C\u002Fstrong>: Execute local business logic alongside built-in sandbox tools\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Credential Refresh\u003C\u002Fstrong>: Rotate access tokens and API keys while preserving filesystem state and installed packages\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Asynchronous Work Without Blocking\u003C\u002Fh2>\n\u003Cp>The core challenge with long-running agent tasks has been reliability: keeping HTTP connections open is fragile. With \u003Ccode>background: true\u003C\u002Fcode>, interactions now run asynchronously on the server. The API returns \u003Cstrong>an ID immediately\u003C\u002Fstrong>, allowing client applications to poll status, stream progress, or reconnect later while the agent continues working remotely.\u003C\u002Fp>\n\u003Ch2>Direct Integration of External Systems\u003C\u002Fh2>\n\u003Cp>Another major addition is \u003Cstrong>Remote MCP Server Integration\u003C\u002Fstrong> (Model Context Protocol). Instead of writing custom proxy middleware, developers can connect Managed Agents directly to remote MCP servers. This means agents in the sandbox can communicate with private databases and internal APIs – while simultaneously leveraging Google Search or code execution. Tools can be mixed and matched flexibly.\u003C\u002Fp>\n\u003Cp>Additionally, \u003Cstrong>Custom Function Calling\u003C\u002Fstrong> allows local business logic to run on the client while built-in tools execute on the server. The API uses "step matching" to route operations correctly.\u003C\u002Fp>\n\u003Ch2>Credentials Refresh, State Persists\u003C\u002Fh2>\n\u003Cp>For production environments, \u003Cstrong>Credential Refresh\u003C\u002Fstrong> is a practical addition. Access tokens and short-lived API keys expire. Developers can now pass updated network configuration with fresh credentials on the next interaction – and the sandbox retains its filesystem state, installed packages, and cloned repositories.\u003C\u002Fp>\n\u003Ch2>What This Means for You\u003C\u002Fh2>\n\u003Cp>For developers and enterprises building production AI agents, these updates address critical gaps: \u003Cstrong>Background Execution\u003C\u002Fstrong> solves timeout issues with long-running tasks. \u003Cstrong>Remote MCP Integration\u003C\u002Fstrong> enables secure connection to legacy systems and internal APIs without workarounds. \u003Cstrong>Credential Refresh\u003C\u002Fstrong> makes multi-tenant and enterprise scenarios viable. These are the building blocks often missing between prototype and production. Whether your team is already experimenting with Gemini agents or planning to start, these capabilities make reliable, asynchronous agent workflows significantly more accessible.\u003C\u002Fp>\n\u003Ch2>Sources\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fblog.google\u002Finnovation-and-ai\u002Ftechnology\u002Fdevelopers-tools\u002Fexpanding-managed-agents-gemini-api\u002F\">Google AI\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fblog.google\u002Finnovation-and-ai\u002Ftechnology\u002Fdevelopers-tools\u002Fexpanding-managed-agents-gemini-api\u002F\">blog.google\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cem>Editorially owned by \u003Ca href=\"\u002Fen\u002Fautor\u002Fideal-syka\">Ideal Syka\u003C\u002Fa>. Sources and method: \u003Ca href=\"\u002Fen\u002Fredaktion\">Newsroom & method\u003C\u002Fa>. Tips and corrections: \u003Ca href=\"mailto:ai@i6eal.de\">ai@i6eal.de\u003C\u002Fa>.\u003C\u002Fem>\u003C\u002Fp>\n",1783503407693]