Compared to other CLI-based AI agent IDEs, Antigravity shows very weak skill awareness. I constantly have to remind it to check and use skills, which wastes time.
I’ve tested Claude Code, Cursor, Codex, and Antigravity on the same project, with the same code and identical prompts. Antigravity Gemini Pro 3.1 is the only one that failed to utilize skills.
This is not a minor inconvenience—it is a fundamental flaw. For an AI IDE in 2026, such poor skill integration is below standard and unacceptable. I strongly urge the development team to prioritize fixing this issue and ensure Antigravity meets the expectations of modern AI tooling.