What it is
A single-user daily research-monitoring pipeline for researchers with standing questions. It fetches sources, screens them per question, and writes question-centric Obsidian notes. Notes are currently written in Japanese.
How it uses Jev
For each (source, question) pair, Jev answers hard gates as yes/no questions, then weighted scores. Code applies thresholds to route pairs to Keep, Review, Drop, or Incomplete. Jev also marks which verbatim sentences advance or contradict the question, and scores Qwen's draft section on a rubric.
Primitives:choicescorenoul
Technique worth stealing
Move judgment into cheap typed questions so the loop is machine-checkable, with raw probability distributions stored per decision.
Try it
uv sync; uv run jrp run
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-centricyes0.91
- Shows a System One patternyes0.89
- Handles uncertaintyno0.38
- Measuredno0.08
- Runnableyes0.70
- Worth recommendingunclear0.59
- Model replicano0.05
- Problem scopescore on a 0–2 scale0.99
- About Jevyes0.97
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 23 Sept 2026.