What it is
Sniff Test reads Markdown and plain text and checks it against house rules, printing one line per flag with file, line, rule and confidence. It is for writers and teams who want to catch AI writing tells.
How it uses Jev
For judgment rules, it sends one paragraph at a time to Jev with the wording of each judgment rule, and gets back one probability per rule. A rule counts as a flag at or above the threshold (default 0.7). Readings between 0.4 and 0.6 are reported as no judgment.
Primitives:choice
Technique worth stealing
Send one paragraph per request with every judgment rule's wording, and treat the 0.4–0.6 band as no judgment.
Try it
npm install --global snifftest; snifftest check --dry-run draft.md
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-centricyes0.61
- Shows a System One patternyes0.82
- Handles uncertaintyyes0.62
- Measuredyes0.97
- Runnableyes0.95
- Worth recommendingunclear0.50
- Model replicano0.10
- Problem scopescore on a 0–2 scale1.11
- About Jevyes0.84
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.