Read .ai-bridge/current-plan.md and execute it in small, reviewable steps.
After each meaningful change, update .ai-bridge/agent-status.md with:
- what changed
- files touched
- tests, lint, or typecheck commands run
- results
- blockers or questions
- what ChatGPT or another reviewer should review next
Keep .ai-bridge/decisions.md aligned with implementation choices. Do not overwrite .ai-bridge/current-plan.md unless explicitly asked.