[ Google IO 2026 ] Antigravity CLI 實際使用筆記與 Gemini CLI 搬家指南
Summary
Google has announced the deprecation of the Gemini CLI, effective from May 19 to June 18, 2026, requiring users to migrate to the new Antigravity CLI (command `agy`). This change is part of Google's strategy to unify its AI development platforms under the "agent-first" Google Antigravity ecosystem, where the Antigravity Harness serves as a shared agent runtime across Gemini API's Managed Agents, AI Studio's coding agents, and Antigravity 2.0 desktop. The new Antigravity CLI, rewritten in Go, offers advantages such as shared agent runtime benefits, including Dynamic Subagents and Scheduled Tasks, and built-in asynchronous, multi-agent orchestration. Migration involves updating CI/CD scripts from `gemini` to `agy` and transferring `~/.gemini/` configurations, with official tools provided to facilitate this process on macOS, Linux, and Windows.
Key takeaway
For AI Engineers and Machine Learning Engineers currently using the Gemini CLI, you must migrate to the Antigravity CLI (`agy`) by June 18, 2026. This transition is not merely a command-line change but an upgrade that integrates your terminal workflow with the broader Antigravity 2.0 agent ecosystem, enabling advanced features like Dynamic Subagents and asynchronous multi-agent orchestration directly from your CLI. Ensure you use the official migration tools to transfer your configurations and update CI/CD pipelines promptly.
Key insights
Google is unifying its AI agent development tools under a single Antigravity CLI and shared runtime.
Principles
- Unified runtimes enhance feature parity.
- Agent-first platforms streamline development.
Method
Install Antigravity CLI via platform-specific scripts, then use the `agy plugin import gemini` command to migrate existing Gemini CLI configurations and extensions.
In practice
- Update CI/CD scripts from `gemini` to `agy`.
- Authorize project folders for agent read/write access.
Topics
- Antigravity CLI
- Gemini CLI Migration
- Google Antigravity
- Agent-first Development
- Antigravity Harness
Best for: AI Engineer, Machine Learning Engineer, Software Engineer
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by LLM on Medium.