Documentation Ai Rules for Claude Code

# Documentation Ai Rules for Claude Code

Use this ruleset in `CLAUDE.md`.

## Universal AI coding principles
- Be explicit with goals, constraints, and acceptance criteria.
- Enforce clean architecture, readable naming, and test coverage.
- Validate inputs, handle errors intentionally, and avoid silent failures.
- Keep changes small, reviewable, and reversible.
- Write docs for non-obvious decisions and trade-offs.

## Prompting and implementation rules
- Ask for clarifications when requirements are ambiguous.
- Prefer deterministic logic and stable interfaces.
- Avoid introducing unneeded dependencies.
- Match existing style/lint/test pipeline exactly.
- Include migration notes for any breaking behavior.

## Shipping checklist
- Lint passes
- Type checks pass
- Tests pass
- Performance and security risks reviewed
- README/changelog updated when behavior changes