Antigravity (Interactions)

Run the Antigravity managed autonomous agent in a sandbox through the Gemini Interactions API.

Run the Antigravity managed agent through the Gemini Interactions API. Antigravity is a general-purpose autonomous agent (Gemini 3.5 Flash) that plans, runs code, browses the web, and produces artifacts (PDFs, HTML, slides) inside a managed Linux sandbox.

The environment parameter selects the sandbox:

Value	Behavior
"remote"	Fresh remote Linux sandbox.
"env_<id>"	Reuse a previously provisioned sandbox.
dict	Full EnvironmentConfig (sources, network rules).

GeminiInteractions in Agno 2.7.4 sends Antigravity requests in the foreground and forces store=True so each interaction is retrievable. Antigravity also runs in the foreground through the Google Gen AI SDK, which rejects background=True for this managed agent. Use streaming to receive progress while the connection stays open.

Code

antigravity.py

from agno.agent import Agent
from agno.models.google import GeminiInteractions

agent = Agent(
    model=GeminiInteractions(
        agent="antigravity-preview-05-2026",
        environment="remote",
    ),
    markdown=True,
)

if __name__ == "__main__":
    agent.print_response("What is the capital of France?")

Usage

1. Set up your virtual environment

   Mac

   uv venv --python 3.12
   source .venv/bin/activate
   

   uv venv --python 3.12
   .venv\Scripts\activate
   
2. Set your API key

   export GOOGLE_API_KEY=xxx
   
3. Install dependencies

   uv pip install -U "google-genai>=2.0" agno
   
4. Run Agent

   Save the code above as antigravity.py, then run:

   python antigravity.py