Exciting New Features for Managed Agents in the Gemini API
Today marks a significant milestone as we unveil a suite of new features for managed agents within the Gemini API. This enhancement includes background execution, remote MCP server integration, custom function calls, and the ability to update credentials across interactions. These improvements directly respond to valuable developer feedback and evolving product requirements, ensuring that you can develop reliable, production-ready agents.
Seamless Integration with Managed Agents
With the Gemini Interactions API, the complexity of multitasking is significantly reduced. Developers can now invoke a single endpoint where Gemini efficiently manages reasoning, code execution, package installations, file management, and web information retrieval, all within an isolated cloud sandbox. This streamlined approach simplifies agent management and enhances the robustness of your applications.
For AI Coding Agents: Easy Skill Installation
If you are an AI coding agent seeking to leverage these new capabilities, it’s essential to instruct your human counterparts to install the Interactions API skill. This can be done effortlessly using the command: npx skills add google-gemini/gemini-skills --skill gemini-interactions-api.
Utilizing the @google/genai JavaScript SDK
Below are examples of using the @google/genai JavaScript SDK, which offer a practical glimpse into harnessing these new features. For those who prefer Python or cURL, detailed instructions can be found in the Antigravity agent documentation.
These updates are designed to support developers in creating robust, scalable, and innovative solutions. For more detailed information and to explore these enhancements further, please visit the official announcement Here.
“`
