system-one-adapter-rust
Rust port of TypeSafe's system-one-adapter that answers System One questions with OpenAI or Anthropic LLMs for comparison with Jev.
Overview
This is a Rust port of the Python system-one-adapter package: a drop-in replacement for TypeSafe's system_one evaluation API backed by LLM APIs instead of TypeSafe, useful for comparing TypeSafe against an LLM on cost, speed and intelligence. Compatible question, answer, retry and error types are implemented locally, so the crate does not depend on a TypeSafe crate.
OpenAI and Anthropic providers are optional features enabled by default. As in the Python version, the client is configured with how the LLM answers (structured outputs, probabilities or discrete answers, normalisation), each call names a provider and model or passes a provider instance such as an OpenAI-compatible endpoint, OpenAI uses the Responses API with strict JSON Schema, and Anthropic has a configurable output token limit with a clear error when reached. Responses mirror the SDK's typed views and add usage totals across retries.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | The book review is positive. | — |
How it works
- cargo add system-one-adapter with the needed features
- Configure the answer mode
- Call system_one with a provider and model
- Compare results with Jev
Strengths
- No dependency on a TypeSafe crate
- Mirrors the Python adapter's behaviour
- Usage tracked across retries
Caveats
- Community port
- LLM answers may not be calibrated
Getting started
cargo add system-one-adapter, then SystemOneAdapterClient::new(true, AnswerMode::Probabilities).system_one(...).At a glance
- Type
- Project
- Status
- beta
- Language
- Rust
- Runtime
- Rust library
- Has code
- yes
Use cases and integrations
How curated lists describe it
Rust port of TypeSafe system-one-adapter (LLM-backed system_one evaluations).
hellogumbo/awesome-jev
Frequently asked questions
What is this?
A Rust port of the adapter that answers System One questions with LLMs.
Does it call Jev?
No; it replaces Jev with an LLM for comparison.
Which providers are included?
OpenAI-compatible and Anthropic.
Related projects
TypeSafe's official drop-in replacement for the SDK's system_one call backed by OpenAI, Anthropic or Gemini, for comparing Jev with LLMs.
LibraryproductionPythonC# System One client and benchmark comparing Microsoft-Decision-1, Jev and open-jev on the same labelled support tickets.
LibraryprototypeC#Small Node TypeScript CLI for trying Jev on synthetic support text: one question, three in one request, and latency benchmarks.
BenchmarkprototypeTypeScriptSmall harness calling Jev via OpenRouter's Decisions API, with 17 doc-derived cases and a latency and cost benchmark against two LLMs.
BenchmarkprototypePythonOne-page playground and test bench for Jev with TypeSafe, OpenRouter, LLM-baseline and mock backends behind one /v1/systemone contract.
ServiceprototypeTypeScriptRust client, CLI, terminal UI and eval harness for asking System One questions of a local model served on Xeon Phi cards.
LibraryresearchRust