What it is
A Markdown knowledge base for users who want to import documents, classify their subject, review uncertain results, and search original text while preserving exact source bytes.
How it uses Jev
Jev classifies each document's primary subject using a Choice question over the source text. The result, probabilities, and metadata are stored as an immutable decision. Low probability or a small gap between top choices triggers review, where a human can record a correction.
Primitives:choice
Technique worth stealing
Record every classification decision with source hash, taxonomy, question, model, probabilities, and token usage for auditability.
Try it
git clone https://github.com/robebots/kb-jev.git && cd kb-jev && npm ci --include=dev --ignore-scripts && npm run demo
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-centricyes0.79
- Shows a System One patternunclear0.58
- Handles uncertaintyno0.10
- Measuredno0.04
- Runnableyes0.82
- Worth recommendingno0.33
- Model replicano0.09
- Problem scopescore on a 0–2 scale1.05
- About Jevyes0.97
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 23 Sept 2026.