What problem does it solve?
It solves the problem of AI generating code that conflicts with your repository’s established naming, structure, and style by learning real conventions from the existing codebase before enforcing anything.
Core Features & Use Cases
- Convention detection across architecture layers: Learns structure patterns using the project graph, then validates type and naming conventions via a public API snapshot.
- Configuration-aware enforcement: Respects explicit rules from .editorconfig and Directory.Build.props instead of imposing generic defaults.
- Consistency-first code review and generation: Flags deviations from the detected conventions and generates new code that matches the team’s existing style.
- Anti-pattern tracking: Monitors recurring quality issues over time and recommends documenting them as conventions when they repeat.
Quick Start
Run the convention learning workflow by asking Claude to detect and summarize your repo’s coding conventions and then apply those conventions when generating the next feature.