10 open-source Jev projects you can copy tonight
From a drone simulator and a stock decision terminal to an MCP decision server, 10 open-source repos that use Jev for structured decisions, each with a GitHub link.
Jev builds are multiplying faster than anyone can keep track of. If you got Jev API access and still haven't shipped anything with it, here are 10 more you can copy tonight: 1. typesafe-jev-drone-demo A drone simulator in the browser. Python owns position, velocity and checkpoints, Jev picks the next flight action with one Choice question. https://github.com/kxzk/typesafe-jev-drone-demo 2. Jevinik A stock decision terminal. Five parallel evidence searches (market data, news, analysts, macro), then Jev gives a probability that the stock trades higher in 30 days. https://github.com/unicodeveloper/jevocks 3. PageGrade A Chrome extension that grades any landing page. Jev scores every section on clarity, writing and SEO, then gives the whole page an A to E grade. https://github.com/kitze/pagegrade 4. Jev Playground Jev plays tic-tac-toe and Connect Four against other evaluation models, live in the browser. https://github.com/hegargarcia/jev-playground 5. jev-scout Finds open-source repos and crates without hallucinating. Pulls real results from GitHub and http://crates.io, then Jev scores fit and maintenance in one ~120ms call. https://github.com/AkashPriyadarshii/jev-scout 6. jev-pref Turns your AGENTS.md preferences into a linter. Jev checks every diff against your own project rules, advisory or blocking. https://github.com/doeixd/jev-pref 7. decide-mcp An MCP server your agent can hand its decisions to. Give it a question, context and options, get back a recommended option with probabilities. https://github.com/dakdevs/decide-mcp 8. typesafe-jev-workflow Email routing with LangGraph. Jev reads the email, answers invoice or general as a typed Choice, and the right handler takes over. https://github.com/GiesN/typesafe-jev-workflow 9. kev A 0.5B Jev-style decision model built on Qwen2.5-0.5B. Answers typed questions in one forward pass, around 160ms on a MacBook. https://github.com/jaredpalmer/kev 10. Laya A non-autoregressive decision engine. 421M parameters, typed answers in one forward pass, 32.8ms on a GPU. https://github.com/NandhaKishorM/laya Copy these Jev blueprints, then read the full Jev setup below ↓ ↓
This post quotes another post: x.com/hanakoxbt/status/2101709924828934222