Skip to content

rotom

Local Rust gateway that exposes Codex, Grok, Kiro or Vercel AI Gateway credentials as OpenAI/Anthropic APIs, including Jev evaluations.

Servicebeta

Overview

rotom is a small local gateway written in Rust. You log in once to providers such as OpenAI/Codex, Grok, Kiro or Vercel AI Gateway, run rotom serve, and point Claude Code, the OpenAI or Anthropic SDKs, or any compatible client at the local address. Every logged-in provider is exposed at once, credentials are stored per provider, and the advertised model list can be overridden in config.

The Jev connection is through Vercel AI Gateway's evaluation models: rotom serves POST /v1/evaluations for models such as typesafe-ai/jev, which return typed answers and probabilities rather than text, and carries them in its model catalog. That lets tools already pointed at rotom reach Jev through the same local endpoint and Vercel credentials.

What the decision model decides

TypeQuestionAnswers
OtherTyped evaluation questions forwarded to typesafe-ai/jev via /v1/evaluations—

How it works

  1. rotom login --provider vercel (and others)
  2. rotom serve on a local address
  3. POST typed questions to /v1/evaluations with model typesafe-ai/jev
  4. Receive typed answers and probabilities

Strengths

  • One local endpoint for several providers
  • Reach Jev with Vercel AI Gateway credentials
  • Works with Claude Code and standard SDKs

Caveats

  • Jev access depends on Vercel AI Gateway
  • Licence not stated in the fetched README

Getting started

cargo install rotom && rotom login && rotom serve --bind 127.0.0.1:14550 --api-key local-secret

At a glance

Type
Service
Status
beta
Language
Rust
Runtime
Local gateway daemon
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

  • Local gateways: OpenAI- and Anthropic-compatible API gateway that carries Jev through its model catalog and evaluation path.

    yibie/awesome-jev

Frequently asked questions

How does rotom expose Jev?

Through POST /v1/evaluations for Vercel AI Gateway evaluation models such as typesafe-ai/jev.

Do I need a TypeSafe key?

No; it uses your Vercel AI Gateway credentials.

Which clients work with it?

Claude Code, the OpenAI and Anthropic SDKs, and other compatible clients.

Related projects

  • Go harness for System One models: hosted Jev, open Kev in-process via llama.cpp, and Cloudflare Clef, with typed helpers and a CLI.

    ServicebetaGo
  • TypeScript library for one-line local decisions with open models (kev-0.6b, kev-4b, open-jev), no API keys or cloud.

    ProjectbetaTypeScript
  • Swift 6 bridge that runs System One decision models behind Apple's Foundation Models API, on device with Laya or via hosted endpoints.

    ProjectbetaSwift
  • PHP library, extracted from LLPhant, for typed Jev Noul, Choice and Score calls, also usable with a Laya endpoint.

    LibrarybetaPHP
  • Unofficial Rust gateway routing /v1/systemone requests to TypeSafe, Vercel, OpenRouter or local Laya via an editable DAG.

    ProxybetaRust
  • Zero-dependency local gateway and live visualiser that proxies Jev calls, logs every decision to SQLite and draws them on a real-time map.

    ProxybetaJavaScript