Community · E2 · artifact verified
Check suspicious messages for scams with typed verdicts
Paste a suspicious SMS, email, DM or listing into ScamCheck - web app, API or browser extension - and get a scam verdict, risk score, plain-English reasons and next steps, with all wording from the project's own templates rather than a model.
01 · Role in the system
What Jev does here
The judgment is typed and calibrated: Jev returns the scam verdict and risk score, while deterministic code runs alongside it - URL analysis and PII masking - so the model sees a cleaned, structured view. Every sentence a user reads is assembled from the project's own templates keyed to the typed answers, which means the model decides and the wording can't drift, hallucinate or leak the pasted text back out. Shipped three ways from one core: a web app, an HTTP API, and a Manifest V3 browser extension, with a smoke-test script against the live judge and screenshots of real verdicts in the repo.
02 · Control boundary
Where Jev sits
Jev as the verdict engine of an anti-scam checker: typed scam verdicts and risk scores over sanitized input, deterministic URL and PII checks alongside, and every user-facing sentence assembled from project-owned templates.
Code owns the loop, permissions, thresholds, validation, and side effects. Jev owns only the bounded judgments described above.
03 · Known limits
What this evidence does not prove
- Consumer-tool scope; verdicts are advisory and not a fraud-reporting channel.
- Template coverage bounds the explanations; novel scam patterns need new templates.
- Created 2026-10-01, 0 stars; no measured false-positive rate yet.
04 · Attribution
Public sources
This is a Community record: the project was published by a third-party community author.
- AkashNaickar ↗Community · github · public · checked 2026-10-02