What it is
A pi extension that exposes TypeSafe (Jev, System One) judgments as five pi tools, letting a model make narrow semantic judgments while your code and users keep control of thresholds, weights, and actions.
How it uses Jev
The five tools (typesafe_noul, typesafe_choice, typesafe_score, typesafe_evaluate, typesafe_ask) POST directly to the System One endpoint with native fetch. Each tool returns a probability or distribution for the model to use; the model chooses the state and question, your code handles thresholds and actions.
Primitives:choicescorenoul
Technique worth stealing
Pin recurring rubrics in a YAML/JSON question file so the model only supplies per-call facts, not the criteria.
Try it
pi install npm:pi-typesafe-jev, then run /typesafe setup inside pi.
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-centricyes0.93
- Shows a System One patternyes0.90
- Handles uncertaintyno0.06
- Measuredno0.04
- Runnableyes0.94
- Worth recommendingunclear0.40
- Model replicano0.15
- Problem scopescore on a 0–2 scale1.24
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.