jevbooks

← All projects

typesafe-ai

Twister915/typesafe-ai

Typed Rust client for TypeSafe's System One evaluation API.

What it is

Brings TypeSafe's System One API into Rust as typed judgments for applications. It supports routing support cases, classifying content, scoring items on ordered rubrics, and judging conditions with probabilities. Provides async and blocking backends with observable retries and inspectable errors.

How it uses Jev

Applications send their existing state as text or structured JSON and attach typed questions: Noul for true/false probability, Choice for fixed labels, Score for ordered rubrics. The model returns calibrated probabilities. Code uses answers to drive its own rules for thresholds, routing, and side effects.

Primitives:choicescorenoul

Technique worth stealing

Batch independent typed questions over shared state, then keep thresholds and side effects in your own code.

Try it

Use cargo doc --open --all-features; see docs/getting-started.md for installation and examples.

View on GitHub

judged by Jevjev-1.13.0

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-centricunclear0.55
  • Shows a System One patternyes0.64
  • Handles uncertaintyno0.05
  • Measuredno0.04
  • Runnableyes0.94
  • Worth recommendingno0.06
  • Model replicano0.04
  • Problem scopescore on a 0–2 scale1.23
  • About Jevyes0.95

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.