What it is
Screen abstracts with two decision APIs and compare disagreements.
What it does: Imports corpora, estimates cost, scores with OpenAI Decisions and/or TypeSafe JEV, and builds offline explorer reports.
Why it belongs in the showcase: Clear hosted JEV Noul integration beside a second decision provider.
Caveats: Research beta; protect API keys.
Evidence
Public MIT repo, created 2026-10-27, 0 stars as of 2026-10-29. README documents JEV_API_KEY / TYPESAFE_API_KEY, provider modes openai|jev|both, concurrent resumable scoring, and examples/jev-request.json using model jev-1.14.0 with Noul relevance questions. ESMO corpus release noted separately; repo access may be required for that download while private.
Caveats
- Beta CLI; throughput and screening accuracy are not guaranteed
- ESMO corpus download may require repo access while the release is private
- Scoring sends abstracts and interest text to selected providers
Unofficial case page. Stars / list position observed for heat; do not treat heat as a quality SLA. Review the upstream README before you depend on it.