Purpose
The compliance determination IS the work: license scope x geography x expiry gates every citation for the DECLARED use (#4.1/#4.2). A restricted or expired study is unusable however strong the stat.
When to use
After trend-compare, before any brief is drafted.
Inputs
compared.json + config/rights-register.json.
Steps
- Run scripts/provenance_check.py --compared compared.json --rights rights-register.json --out governed.json
- Quote the record verbatim; excluded studies are named with their restriction so the reader knows what was left out and why.
Output
governed.json - the {provenance_record} hop.
Grounding requirements
Every entry cites the rights register row.
Constraints
- The engine decides; the model never launders a restricted stat by paraphrase - a reworded number is still the licensed number.
- Approving new rights is the licensing team's work, not this plugin's.
Escalation / uncertainty
Missing register rows -> unusable until registered; escalate to research ops.