How to use Agent Manager in Antigravity?
I am going back over how to create multiple agents in Google Antigravity. I covered this earlier in a different series, but a few folks still had trouble with the steps. This clears it up with a slower walkthrough inside your workspace.
Last time I spun up agents in Playground to isolate them from regular code. That kept the agents from touching app code while I tested ideas. Today I will do it inside your workspace so everything lives in namespaces.
Need to build complex AI agents?
My team helps CTOs and founders build autonomous agents with OpenClaw and Antigravity.
Open Agent Manager in Antigravity
When you open Google Antigravity, you might land in the traditional IDE view. If you are not in Agent Manager, click the button labeled Open Agent Manager. That brings you into the agent-focused view.
Open a workspace
Inside the Agent Manager:
- Click Open Workspace.
- Click Open New Workspace if you don't already have one active.
- Name your workspace based on your project (e.g.,
marketing-agents).
Start conversations
Once the workspace is open, you can start a conversation. This is where you configure your agent's initial prompt and context.
Create the first agent
- Click New Agent.
- Give it a specific role, like
Research Assistant. - Provide its system prompt and connect it to your workspace files.
Add more agents
To add more agents, simply repeat the process. They will live inside the same workspace but maintain their own context and conversation history. You can switch between them using the sidebar.
Manage notifications
As you use Agent Manager in Antigravity, you might get notifications when agents complete background tasks. You can manage these in the Settings > Notifications menu to avoid getting overwhelmed.
Final thoughts
Using Agent Manager directly in your workspace instead of the Playground allows you to keep everything organized in namespaces and closer to your actual codebase. Try spinning up a few agents today and see how it streamlines your workflow!
Matteo Giardino is a software engineer and AI consultant specializing in autonomous agents and local AI deployment.