How to migrate from Gemini CLI to Antigravity CLI in 3 steps

Google's deprecation of Gemini CLI with no overlap period means you need to act today. Antigravity CLI is not a drop-in replacement—it's a fundamentally different architecture built around persistent, stateful agents rather than stateless chat completions. Here's how to migrate without losing your custom skills and workflows.

Step 1: Install Antigravity CLI and run the built-in migration tool. Google has provided a migration utility that scans your existing Gemini CLI configuration, including custom skills, hooks, and extensions. It will map these to Antigravity's plugin architecture. Note that not all features map one-to-one—Antigravity does not have a direct equivalent for Gemini's 'inline chat' mode, so you'll need to refactor those workflows as agent skills.

Step 2: Audit your skills and hooks. Antigravity uses a skill-based architecture where each skill is a self-contained module with defined inputs, outputs, and tools. If you had complex hooks in Gemini CLI, you'll need to decompose them into smaller, composable skills. Antigravity also introduces 'subagents'—nested agents that can be called from a parent agent. Use this pattern to replace deeply nested hooks.

Step 3: Test your workflows in Antigravity's sandbox mode. Antigravity includes a sandboxed execution environment that lets you test agent workflows locally before deploying. Run your migrated skills through the same test suites you used for Gemini CLI. Pay special attention to error handling and state persistence—Antigravity's durability model is different and may surface edge cases you didn't hit before.

The key insight is that Antigravity is not just a CLI tool—it's a platform for building agentic workflows. Embrace the shift from 'ask a question, get code' to 'define an agent that writes and tests code autonomously.' If you invest in learning Antigravity's model now, you'll be ahead of the curve as Google doubles down on agent-first development.