API
Machine-readable critiques
Every critique is a typed object rendered to JSON: claim inventory, evidence map, 0–5 scores, severity and confidence, the AI agents used, the human-editor approval, the author-response status, the transparency record and the full version history. The same object renders the human-readable page, so they cannot drift. Critique text is licensed CC BY 4.0.
Endpoints
- GET
/critique/api/reviewsIndex of AGISS evidence reviews (scoping reviews / evidence-gap-maps).
- GET
/critique/api/reviews/{slug}Full evidence review: coverage matrix, gaps, fragilities, codings with verbatim rationale spans, provenance hashes.
- GET
/critique/critiques.jsonCompact index of every published per-paper critique (friendly alias).
- GET
/critique/api/critiquesCompact index of every published per-paper critique.
- GET
/critique/api/critiques/{slug}Full structured per-paper critique object.
Try it
The index lists every critique with a json_url pointing at its full object:
curl -s https://policywindow.org/critique/api/critiques | jq '.critiques[] | {slug, severity, target_journal, json_url}'