What it is
A small, extensible decision-to-action harness for TypeSafe Jev. It connects evidence, Jev judgments, permitted actions, and verified outcomes for one bounded run. Domain-independent core with pluggable data sources and tools. For developers building agent-like loops with Jev.
How it uses Jev
The domain's questions hook returns a map of Choice, Score, and/or Noul questions. The harness sends a batch to an interchangeable evaluator (Jev by default). Answers are validated, then the domain's decide hook combines them into an action or stop reason. Jev does not generate arbitrary commands or tool arguments.
Primitives:choicescorenoul
Technique worth stealing
Batch independent questions in one request; ask dependent questions on the next iteration after a tool result.
Try it
npm run demo; npm run demo -- --live (requires TYPESAFE_API_KEY)
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-centricyes0.88
- Shows a System One patternyes0.61
- Handles uncertaintyno0.07
- Measuredyes0.70
- Runnableyes0.91
- Worth recommendingunclear0.46
- Model replicano0.05
- Problem scopescore on a 0–2 scale1.09
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.