Autonomous Git Sync: Agents That Observe, Classify, and Act — With a Restore Point
TestRelic Team
A couple of weeks ago, Studio and the CLI got a real Git surface — merge, rebase, resolve conflicts, restore points with undo — replacing the old fast-forward-only sync button. That was the foundation. This release builds an autonomous layer on top of it.
Observe, classify, act
Turn on autonomous Git sync for a repository, and an agent watches it, classifies what's changed, and acts — one round trip, without you driving each step by hand. It's opt-in per repo, not a global switch, and every run leaves a restore point, so turning it on is never a one-way door.
Agents propose; they don't get unilateral force
The rule that shipped alongside full Git support still holds, and now matters more: AI agents can propose changes through the Git surface, but they are never allowed to force-push, reset, or discard on their own. This release adds a visible approval surface specifically for agent-run Git actions, so an autonomous sync's proposed action is something you see and approve — not something that happens silently in the background.
Why this is different from "auto-commit"
An agent that watches a repo and commits automatically is a hazard, not a feature — it can't tell you why a change is safe, and it can't be trusted with anything it can't take back. Observe → classify → act is deliberately three distinct steps, and the restore point is the thing that makes step three reversible even when the classification was wrong.
Get started
Enable Git sync per repository from Studio, or run /git ... from the terminal. See Git integration for the full command reference.