10 open-source projects built with Jev on GitHub
A roundup of projects that appeared on GitHub after Jev's launch, including Mac computer control, a browser agent, a Monad trading bot, codebase question answering, a drone, log classification and prompt-injection detection.
Jev has been blowing up on GitHub since launch, here’s what people are already building with it if you have API access and don’t know where to start, copy this: typesafe-computer-use: tell your Mac what you want in plain English, it reads the screen and picks every next click for about $0.0002 a step https://github.com/awlevin/typesafe-computer-use jev-ultrafast: a browser agent where Jev chooses every click, and a text model only wakes up when something has to be typed. 16,758 stars https://github.com/browser-use/jev-ultrafast jev-trader: a live trading bot placing limit orders on Monad every 300ms block, with Jev as the only judge. 1,911 stars https://github.com/jarrodwatts/jev-trader every: asks a yes/no question about every function in your codebase and ranks them by probability. grep, but the pattern is a question https://github.com/sufianetaouil/every jev-drone: a simulated quadrotor flies a five-station obstacle course using only its camera, and Jev reads the situation twice a second https://github.com/RomanSlack/jev-drone commit-miner: sorts your Git commits into bug fixes, security fixes mapped to CWEs, and change types https://github.com/devanshbatham/commit-miner pkg-gate: checks what an npm package’s install scripts do before they run on your machine https://github.com/hemanth/pkg-gate Jev-Moderation-Bot: a Discord bot that catches phishing and social engineering in real time, then escalates the penalty step by step https://github.com/brainstormity/Jev-Moderation-Bot jevlogs: scores every OpenTelemetry log for urgency, so only the ones that matter ever reach an expensive model https://github.com/reachjalil/jevlogs jevegis: prompt injection, jailbreaks and data leaks checked in one API call. MIT https://github.com/0xArx/jevegis in all ten, the model doesn’t write anything. it gets a question and picks from answers someone defined first the full setup is in the article, run the three-question test before you build number eleven
This post quotes another post: x.com/imryven/status/2104255840819491030