What it is
A standalone Rust CLI for developers running agents with large skill libraries. It ranks which skills fit the next step using live session context, and can abstain when none fit.
How it uses Jev
Jev compares candidate skills broadly, then reads richer excerpts from a shortlist and evaluates whether each fits, including a real 'none of these' option. Results are advisory: the agent decides what to consult.
Primitives:choicenoul
Technique worth stealing
Two-stage ranking: broad Choice comparison, then detailed reranking with bounded excerpts, plus sentinel-based abstention.
Try it
sr demo --case useful (offline fixture, no key needed)
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 patternunclear0.58
- Handles uncertaintyno0.14
- Measuredno0.05
- Runnableyes0.97
- Worth recommendingno0.36
- Model replicano0.04
- Problem scopescore on a 0–2 scale1.27
- About Jevyes0.97
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 21 Sept 2026.