OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Frameworks

Google Antigravity SDK

Google Antigravity SDK is a Python library for building stateful AI agents powered by Antigravity and Gemini, with abstractions for the agentic loop and conversation handling.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Google Antigravity SDK — OpenViST project profile in Agent Frameworks.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when developing a Python agent application that benefits from a stateful runtime, built-in conversational abstractions, streaming responses, or multimodal input handling.

Who it is for

  • Python developers building conversational AI agents.
  • Teams integrating agents with Gemini or Gemini Enterprise Agent Platform.

Documented uses

  • Create an agent that accepts prompts and returns conversational responses.
  • Stream response text or inspect conversation steps and accumulated session history.
  • Pass images, video, audio, and documents alongside textual instructions.
Before you choose

Requirements & limits

The SDK is distributed as a Python package on PyPI. Its platform-specific wheels include a compiled runtime binary; the repository says cloning alone is insufficient. Listed platforms are Linux, macOS, and Windows.

  • The SDK defaults to read-only mode; enabling all tools, including writes, requires configuring capabilities.
  • Gemini Enterprise Agent Platform has distinct API-key Express Mode and project/location Standard Mode authentication paths.
  • The higher-level Agent manages lifecycle concerns, while Conversation exposes session history, turn counts, and step-level streaming.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the SDK and run the hello-world example

Install the package with its runtime binary and start the documented quickstart example.

Before you start

  • Python environment
  • Gemini API key
  • Install from PyPI to obtain the platform-specific compiled runtime binary

Steps

  1. Install the package with the command below.

    Documented command
    pip install google-antigravity

  2. Set the GEMINI_API_KEY environment variable to your API key.

  3. Run the command below from the repository.

    Documented command
    python ./examples/getting_started/hello_world.py

Expected outcome

The documented hello-world agent example runs using the Gemini API key.

Guide limitations

  • Cloning the repository alone does not provide the compiled runtime binary needed to run the SDK.
Documentation 1 ↗
Source-based practical guide

Configure the SDK for Gemini Enterprise Agent Platform

Choose and configure one of the two documented authentication modes for a Vertex-backed agent.

Before you start

  • For Express Mode: an API key.
  • For Standard Mode: a Google Cloud project, location and local Application Default Credentials.

Steps

  1. Set vertex=True in LocalAgentConfig.

  2. For Express Mode, provide api_key; for Standard Mode, provide project and location.

  3. Alternatively, set GOOGLE_GENAI_USE_VERTEXAI or GOOGLE_GENAI_USE_ENTERPRISE, along with GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_LOCATION.

  4. For Standard Mode, authenticate locally with gcloud auth application-default login.

Expected outcome

A `LocalAgentConfig` is set up to route agent requests through the selected Gemini Enterprise Agent Platform authentication mode.

Guide limitations

  • Standard Mode requires local Application Default Credentials.
  • Explicit configuration keyword arguments take precedence over environment variables.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
9 October 2026
Original website title
google-antigravity/antigravity-sdk-python · Official homepage title · retained in its source language
How we review sourcesSuggest a correction