Hi everyone,
I’m currently developing a web project using Antigravity IDE—specifically within a “legacy Spring” environment (and yes, it’s just as “classic” and painful as you’re imagining).
To reclaim some of my time, I’m writing a Skill to automate functional testing. The goal is to hand off the testing to Antigravity so I can focus on other tasks. However, I’ve hit a roadblock: every time the Skill attempts to execute JavaScript, the IDE pauses and waits for my manual confirmation.
I fully understand the security implications, which is why I keep my global Review Policy and Terminal Command Auto Execution set to “Request Review.” But for this specific testing Skill, the JS logic is pre-verified and safe. I need it to run autonomously. I’ve tried adding instructions like “execute all commands immediately without asking” to the Skill’s prompt, but the IDE’s system-level safety guardrails (rightfully) override them.
I’ve checked out some external extensions like auto-all-Antigravity, but they’re a bit too “all-or-nothing.” They tend to confirm everything—including undo prompts—and either only target the currently focused window or blast through every chat history at once.
My question is: Is there an IDE-native setting or a more granular extension that allows me to auto-confirm executions only for a specific conversation or Skill session? I want to keep my global security settings tight while giving this specific test runner a “green light” to proceed.
Any advice or clever workarounds from fellow Antigravity users would be greatly appreciated!