Make the decisions explicit
AI Agent Rules is a worksheet for recording the boundaries you want your assistants to follow. The Builder and Personal tracks each contain 100 questions, with ALLOW, ASK, DENY and Doesn’t apply choices.
You can add scope and notes to each decision, review your answers and download a policy. The Builder track also offers settings for supported coding tools, with an explanation of what those settings can and cannot cover.
Review before allowing
Suggested answers are marked as unreviewed until you make a decision. An unreviewed suggested ALLOW is treated as ASK in exported policies and tool settings. That gives you a starting point without quietly granting permission.
Answers can be saved to a file and loaded later. The worksheet runs in the browser and stores answers locally; it can also be downloaded for offline use.
A policy is a starting point
Written boundaries make your expectations clearer. They do not guarantee that every assistant will follow them, and tool settings only cover part of the policy. Those limits are explained alongside the exports.
The project is open source. You can try the worksheet, inspect the implementation and follow the repository for updates.