ai-rules

Why we included this project

Coding assistants are good at producing code that compiles, less good at producing code that fits how your repository is actually layered, and this CLI works on that gap before code reaches review or merge. It reads your Markdown project rules and compiles them into structured rule metadata, stable evidence IDs, and audit and fix prompts that agents like Cursor, Windsurf, and Claude Code can act on directly. The appeal for a tech lead is that conventions get encoded once as a local operating contract, with severity thresholds, exceptions, and path aliases explicit, so you stop re-arguing the same layering violations every sprint and non-standard layouts work without editing every rule. It can also emit a separate business-logic inspection prompt that flags missing authorization checks or unsafe state transitions. If you have ever watched an agent call the data layer from a UI component, this is worth a look before your next review cycle.