Start with your next task.

Your coding agent writes and changes code. Auto Moniter coordinates the work around it: clarifying requirements, scheduling independent reviews, feeding back gaps, and running your acceptance checks until the requested work is verified.

Yes. Auto Moniter works with several native Agent CLIs and selected backends through Paseo. Support and permission modes vary between worker and reviewer roles. Install and authenticate each CLI, then consult the project documentation and local environment checks for the combination you plan to use.

With independent review enabled, the worker and reviewer must agree on the delivery scope, provide valid evidence, and pass the configured checks and version checks. The watcher makes the final acceptance decision. An agent saying “done,” or a formatting check passing, is not enough.

The supervisor and console run locally. Your chosen agents may send code, prompts, and context to a model provider, depending on their configuration. Local orchestration does not, by itself, make model calls offline or keep all data on your machine.

The supervisor saves the workspace, checkpoints, and recorded review feedback. If recovery checks pass, you can resume the same run with its accumulated time and call usage intact. If recovery is unsafe, it stops with diagnostics. Unaccepted work is never reported as delivered.