Observed arrival · 2026-09-18
Safer with Jev: yes/no judgments for code and text
An interactive showcase for Jev, a claimed AI model class that returns binary judgments through a public API.
- For
- Developers screening code, text, and untrusted prompts
- Worth noticing
- The public demos include a curl-accessible /ask-jev endpoint and a separate /nice-try prompt-injection example.
Field notes
Jev is presented as a binary judgment layer rather than a conversational assistant. The homepage supplies examples for TypeScript review, AI-slop detection, prompt-injection screening, policy judgments, and comment moderation, with curl requests targeting public API routes. Its code sample uses a deliberately messy customer function, making the demonstration about criteria-based inspection rather than open-ended generation. The site also directs developers to AGENTS.md for the request contract, while performance and cost improvements remain stated claims.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue