What it does
Its production backend validates all three Choice distributions and records the rubric hash before deterministic policy makes the final call.
Agent tooling
Any Auto asks Jev whether coding-agent tool calls are risky, authorized, and policy-compliant before approving them.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
Its production backend validates all three Choice distributions and records the rubric hash before deterministic policy makes the final call.
Write a list of commands your coding assistant must never run alone, such as deleting files or publishing code. A developer can then install any-auto on macOS or Linux and connect it to your assistant tool.
Whenever the assistant wants to run an action, any-auto checks if it looks safe and follows your rules. If an unexpected error happens or the review takes too long, the program stops the command automatically.