---
goal: Every day, the PLAYBOOK is refined from the previous day's journal
---
Recurring goal (never archived). Once per day — when today's date (the `today`
sensor reading) is newer than the last-improved date below — make ONE small
improvement:
- Read the last ~24h of journal.md. Find ONE recurring mistake, hesitation, or
human intervention and turn it into a clearer PLAYBOOK rule (or sharpen a vague
one). Keep it to a single small change.
- PLAYBOOK changes affect every future move, so PROPOSE the diff to me here and
apply it with `looop _ playbook write` only after I confirm.
- When done, set last-improved below to today's date (via `looop _ goal write
playbook-daily`).
last-improved: never