What it is
A prototype intrusion detection system for network security. It takes a network flow's metadata and returns whether it is an attack and its threat category, with probabilities. Tested on the NSL-KDD benchmark against LLM and machine-learning baselines.
How it uses Jev
Jev receives one network flow plus five labeled examples per category and answers two questions: whether the flow is an attack (Choice) and its category (Choice), returning probabilities and confidence. The p_attack ≥ 0.5 cut decides the verdict; no text is parsed.
Primitives:choice
Technique worth stealing
One request, two answers: both questions run over the same state and come back as a probability, an option and a confidence.
Try it
uv run jev-ids run --dataset data/nsl-kdd/dataset.json --detector jev --split smoke --k 0,1
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.85
- Shows a System One patternno0.09
- Handles uncertaintyno0.04
- Measuredyes0.97
- Runnableno0.33
- Worth recommendingunclear0.49
- Model replicano0.07
- Problem scopescore on a 0–2 scale1.54
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 23 Sept 2026.