CodeSync Use case
Code review automation
What gets automated
A review run scores the snapshot from the diagnose findings and lists next actions. If Workers AI or your own key answers, a model note is stored. If the model errors, the error is stored and the score still comes from the findings.
Expert can schedule a weekly watchdog, which repeats the diagnose. Extreme can schedule it daily and can run a mission: diagnose, safe repair, review, and generate in one job. Extreme can also open a GitHub pull request when the safe repair changed files.
What you still review
Generated files and pull requests are output for a human to read. The terms of service say you are responsible for merging them. The product does not auto-merge.