# AWR-004 offline decision lab

1. Pick a fixture from `fixtures/FIXTURE_BINDINGS.json`.
2. Copy its exact `microservices` and `full_offline_studio` evidence into the two records of `evidence/decision_record.template.json`.
3. Choose a visible decision state and write your own rationale, review due date, and stop/pivot trigger.
4. Run `python run_checks.py <submission.json>`.

The checker has no hidden expected verdict. It checks exact shipped evidence binding and visible consistency rules only. All data are synthetic.