Teaching the AI assistant the house rules
AI coding assistants are powerful and opinionless: out of the box they'll happily produce monolithic files, skipped error paths, and whatever style the prompt implies. This project is a set of custom Claude Code skills and CLI tooling that encode real production standards, so AI-assisted development follows them automatically.
- Type
- Developer tooling · AI workflow
- Stack
- Python · Claude Code · CLI
- Form
- Custom skills + command-line tooling
- Goal
- Production standards enforced in AI-assisted work
The problem
An AI assistant without context regresses to averages — average file structure, average error handling, average naming. Teams that adopt AI coding tools without encoding their standards spend the saved time reviewing avoidable mistakes. The fix isn't better prompting in the moment; it's packaging the standards once so every session starts from them.
What it is
A collection of custom skills for Claude Code — rule sets that load automatically when relevant work begins, covering things like code structure, framework conventions, and quality gates — plus CLI tooling that streamlines the developer workflow around them. The effect is that the assistant behaves like a team member who has read the engineering handbook, because in a real sense it has.
Notes
Built in Python around the Claude Code extension points. This portfolio — and the data-driven architecture behind its case studies, schema, and llms.txt — is itself developed under these skills.