Code Style Consistency

Problem It Solves

Models often generate code with default habits: a project uses @/ aliases while output writes ../; a project wraps request, but generated code calls fetch; a project uses cn() for Tailwind classes, while output writes conflicting raw classes. The code may run, but it feels like a temporary contribution. This skill focuses on identifying project conventions before generation.

How It Works

• Style fingerprinting: uses scripts/scan_style.py or user-provided reference files to extract stack, naming, import style, component patterns, CSS approach, request wrappers, and error handling.
• Reference-first generation: before adding components, pages, or APIs, it locates the closest existing file and mirrors its skeleton, props, imports, and exports, replacing only business logic.
• Hard constraints: keeps UI libraries, styling, state management, TypeScript strictness, naming, and comment language aligned with the project.
• Explicit deviations: if a generated piece must depart from project style, it states the reason instead of silently changing conventions.

Boundaries

Best for frontend projects, existing codebases, and clear tech stacks; the source materials specifically mention scan_style.py, Ant Design, Tailwind, and CSS Modules. If conventions exist only in tribal knowledge, the user still needs to provide reference files or answer a few probe questions.