Antigravity code reveals Gemini-powered Vetted mode
Version 2.12.2 contains a gated permission preset that assigns a second AI system to approve or reject an agent's proposed actions.
By Ryan Merket · Published
RUNTIMEWIRE INVESTIGATION — Scoop
Original reporting by RuntimeWire, based on reverse engineering.
Why it matters
Vetted would let developers trade some manual approval prompts for a second model's judgment. Its value depends on rollout scope, failure defaults and measured security performance that the distributed code does not establish.
Reporting record
Finding
Antigravity 2.12.2 contains a gated permission preset backed by a security evaluator called Policy Guardian. Recovered code and prompts describe checks on tool calls, downloads, dependencies and attempts to repeat denied actions.
How we verified
Methods: reverse engineering.
The Vetted beta option exists in the 2.12.2 UI, literal preset definition and settings dropdown wiring
Reverse engineered the package.
Tested versions: Antigravity 2.12.2.
Reproduction
RuntimeWire partially reproduced the finding.
File hashes
language_server.exe: 424ed4ef6fc9cb470ce785d8f4b392de3fa94fd16d030e6ff83e043440659f94app.asar: a75535fec1f039be8e78c6a66f3e41f15969101fcf44a0690b7dce4d998b875b
Company response
The company did not respond to requests for comment.

Google Antigravity has built a permission mode that assigns Gemini a supervisory role over commands an AI coding agent wants to run, according to RuntimeWire's static examination of files distributed with Antigravity 2.12.2.

The setting is labeled "Vetted (beta)." Its interface description says: "A Gemini-powered security agent decides if commands should be auto-approved." The option is controlled by an eligibility flag, leaving its availability to users dependent on a server-supplied experiment configuration. RuntimeWire did not activate the setting through a live account.