Google Antigravity AI Agent Expands to VS Code, JetBrains, Zed

Google is bringing its Antigravity AI coding agent directly into developers' existing code editors, launching extensions for Visual Studio Code, JetBrains IDEs, and Zed alongside new enterprise management capabilities through Gemini Enterprise.

The move addresses a fundamental friction point: when Google launched Antigravity in November 2025, developers had to work inside a separate desktop application. The new extensions let developers open agent conversations in a side panel, review code changes inline, and delegate multi-step tasks without leaving their primary editor.

Why it matters

Meeting developers in their existing workflows removes a significant adoption barrier for enterprise teams, where standardizing on a single new tool is often impractical. More importantly, the Gemini Enterprise integration gives IT administrators the budget controls and access policies they need to deploy autonomous agents at scale—a capability that has become critical as organizations discover how quickly AI coding costs can spiral.

Cross-editor consistency

The VS Code extension is available now through Microsoft's marketplace on macOS, Linux, and Windows. The Visual Studio 2026 extension for .NET solutions is in preview, while JetBrains support covers IntelliJ IDEA, PyCharm, WebStorm, GoLand, CLion, and Rider starting with version 2026.2.1. Zed is also supported.

Developers use the same Antigravity account across all environments, eliminating separate sign-ins or license management. Agent sessions inherit the organization's IAM policies, VPC Service Controls, and regional data boundaries regardless of which editor initiated them.

Enterprise budget controls arrive

Administrators can now enable Antigravity for employees on Gemini Enterprise Standard, Plus, or Standard Emerging Market plans. They can set monthly spending caps for each Google Cloud project—when the included quota runs out, Antigravity either stops or switches to pay-as-you-go pricing.

Google meters the allowance using a rolling seven-day pool, with unused credits expiring at reset. Currently, everyone in the same edition, project, and region draws from a single pool. Controls for individual users and teams are planned for later this year.

Token consumption adds up fast

The budget controls address a real problem. A nontrivial engineering task can consume 150,000 to 200,000 tokens, while multi-agent handoffs add more input tokens each time work passes between agents. One built-in Claude Code skill was recently found to be loading more than 200,000 tokens before answering a question.

Without per-developer allocations, an agent-heavy workflow could exhaust a team's entire pool within hours. Microsoft recently introduced AI token budgets for its engineering divisions after discovering many engineers were spending hundreds to thousands of dollars monthly on tokens. Uber reportedly exhausted its entire 2026 AI coding budget in the first four months of the year, while an internal Amazon project exceeded its planned budget by 860%.

Security and access management

Enterprise developers authenticate with company credentials, then select the Google Cloud project and region for Antigravity to use. Organizations can connect to existing identity providers via Workforce Identity Federation, or developers can authenticate using Application Default Credentials.

Administrators can limit an agent's workspace, block browser access, and decide which MCP servers it can connect to. Google notes that data from enterprise Antigravity sessions is not used to train its foundation models.

The extension strategy lets Google deploy agents onto developers' machines without requiring a company-wide tooling change, while the Gemini Enterprise integration ensures every session follows the same policies and project budget regardless of editor choice.

These details were first reported by The New Stack.