Independent research repository on Jev: sourced findings that separate verified claims from vendor marketing repeated by others, a rubric-scored catalog of public Jev repositories, and recurring design patterns with evidence.
User repository on GitHub
Independent, source-linked research on TypeSafe AI's Jev (System One), with 947 rubric-scored public repositories, recurring patterns, datasets, and bilingual documentation.
Repo stats from the GitHub API, cached Oct 2026.Homepage
A sourced, critical research file on Jev — TypeSafe AI's "System One" decision model — plus a rubric-scored map of what people are actually building with it. Jev does not write text. You give it a state and typed questions (choice, score, noul = yes/no); it returns probabilities. That makes it cheap, fast and easy to wire into software — and also easy to over-sell. This repository tries to answer three questions with evidence instead of enthusiasm:
Quoted from the project's README. Read the full README.
Watch out
Scores are an LLM's judgment against a written rubric rather than benchmarks of the projects themselves, and for most repositories the author read the description, metadata and the first 12,000 characters of the README instead of the code. "Calls Jev" only means the README or code shows a real call, and the jev GitHub topic is free-form, so many tagged repositories are unrelated and get marked as such rather than dropped. The README also notes that Jev is not deterministic call to call and that vendor accuracy figures measure agreement with judge models, not ground truth. Entries for deleted repositories are kept and marked gone. Copy reviewed against the repository README on 2026-10-04, not from running it.
erendikmenn/jev-rag-benchmarkPython
amithgc/local-jevPython
lgy1027/jevshieldPython
wangkuangkuang/jev-mcp-serverPython
shamazharikh/qwen-rlcdPythonOriginal write-ups on what this entry is for.
Posts from builders shipping with Jev right now.
