31 lines (22 loc) · 922 Bytes

Review Checklist

Run before every push. Keep it short.

Before Push

npm test passes
npm run build passes
Every new function has a purpose that aligns with the goal
No dead code, no commented-out blocks
Error messages are helpful (not just "error")
README is updated if user-facing behavior changed

Goal Alignment Test

Every PR must answer YES to at least one:

Does this help a developer understand what their AI did?
Does this reduce friction in the review workflow?
Does this make session context searchable or retrievable?
Does this reduce the gap between "AI wrote code" and "human approved it"?

If none, the feature is out of scope.

Code Standards

Single responsibility per function
No magic strings (use constants or config)
Types over any
Prefer pure functions where possible
Tests for new logic (1-3 tests per module, not exhaustive)