Connect Google Antigravity

3 minute read · Chats & Local Agents

Connect Google Antigravity once, then choose it in a project conversation. Your Vortex sign-in and provider sign-in are separate accounts.

• Connect the account
• What to expect from Google Antigravity
• Check installation, account and model separately
• Read the Google usage display correctly
• Work through an approval or interruption

Connect the account#

1. Open Settings → Agents & chat and select Google Antigravity.
2. Check installation status. If needed, choose Install and follow preparing, installing and verification progress. An error or cancelled installation needs attention before a run can start.
3. Choose Sign in. Complete the supported browser flow and any displayed provider code, then return to Vortex to verify the account status.
4. Start a new chat, choose Google Antigravity and an available model, then ask it to explain the first step of your app idea.

What to expect from Google Antigravity#

Complete the current Google account connection flow. Account eligibility can differ from having the runtime installed. Its usage reports describe per-model quotas and reset times; this integration does not invent a conversation token count.

If the app reports unknown or signed out, open the connection controls again. An unknown state is not a confirmed sign-in. A provider account rejection needs an eligible account or another supported agent; Vortex does not change provider eligibility or include its tokens.

Continue with your first app, model and usage choices and permissions.

Check installation, account and model separately#

Use the Google Antigravity entry in Agents & chat. Follow the installation progress, then launch the supported sign-in flow and return for connection verification. Cancel through the dialog if you need to abandon an attempt.

Open a new project chat, choose an available Google model and ask it to inspect the project's setup. A successful response that identifies the selected checkout is a stronger connection check than an executable being present. If the account is rejected, inspect the provider's eligibility message and use an eligible account; signing in to Vortex does not grant access to a Google model.

Read the Google usage display correctly#

This integration reports per-model account quota and reset times when supplied. It does not report conversation token counts. A blank conversation-token display is therefore different from an empty account allowance. Refresh the popover to inspect the currently available model limits, and do not use a reset-expired sample as evidence that a new run is blocked.

Model and reasoning choices vary. Choose from the supported options shown in Vortex rather than assuming a setting available for Codex or Claude has the same meaning here.

Work through an approval or interruption#

Begin with Supervised access for an unfamiliar project. When an action needs approval, review the project, command or tool details and answer in the owning conversation. If you stop the run, its old approval cannot be reused for a later run. Additional instructions may be delivered as steering or held for the next turn; read the queue state before repeating them.

For support, capture the installation/connection stage, selected model and short provider error. Remove account identifiers, private paths and tokens from shared material. Continue with usage and models.