# AWR-006-LAB1 — Offline Decision Fixture

Goal: construct `Portfolio/Creator Marketplace Control Decision Record v1` from a shipped synthetic fixture and defend two independent decisions: `PORTFOLIO_CONTROL` and `CREATOR_MARKETPLACE`.

1. Choose one fixture listed in `fixtures/FIXTURE_BINDINGS.json`.
2. Copy its two evidence objects exactly into a copy of `evidence/decision_record.template.json`. Each evidence object includes `problem` and `evidence_ref`; the Portfolio axis uses the canonical field name `revision_id`.
3. Author `decision_state`, `rationale`, `review_due`, and `stop_or_pivot_trigger` for each axis. `rationale` must be non-empty and learner-authored. Do not change fixture evidence.
4. Run `python run_checks.py <your-submission.json>`.
5. Repeat with the negative and transfer fixtures. A missing decisive fact must stay a typed non-success state; the checker never supplies an expected verdict.

For the marketplace axis, a `PILOT` may be a bounded non-transactional directory/referral experiment when the selected fixture explicitly sets a non-transactional `marketplace_role`; transactional gates apply only to a proposed transactional role.

Visible state vocabulary: `USE`, `PILOT`, `NOT_NOW`, `UNKNOWN`, `EVIDENCE_INCOMPLETE`, `OWNER_MISSING`, `REVIEW_REQUIRED`, `HOLD`, `PARTIAL`, `DISPUTED`.

The lab is fully synthetic and offline. It requires no provider account, credentials, production data, payments, payouts, publication, or live service mutation. The rules in `CHECK_RULES.json` are the complete learner-visible consistency contract; there is no hidden answer key.