convention-learner

Coding conventions are detected by analyzing project structure, naming patterns, and layering across the codebase. The skill examines representative public types and existing folders to infer actual style norms rather than relying on generic templates.

Project-specific naming patterns are enforced by learning existing conventions and tracking deviations during code review. The skill identifies likely convention violations and suggests aligning generated code to the discovered patterns.

Convention detection respects explicit configuration sources like .editorconfig, Directory.Build.props, and global.json. These project rules act as authoritative constraints that override any generic default style enforcement.

The best way to document coding conventions is generating a structured Detected Conventions summary. This evidence-based output captures project patterns and anti-patterns to persist knowledge for onboarding and code generation.

Anti-patterns and convention violations are tracked over time using recurring quality issue detection. The skill flags deviations during generation and review to prioritize lasting convention improvements across the project.

Analyzing existing codebase patterns prevents slow, inconsistent development by turning actual conventions into actionable guidance. Standard templates guess rules blindly, while evidence-based detection enforces true architecture consistency.