Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

taurus-jev-sdk-go

Unofficial dependency-free Go client for TypeSafe's System One API with strict response validation, masked keys and bounded retries.

LibrarybetaMIT
View on GitHub ↗

Overview

taurus-jev-sdk-go is an unofficial Go client for TypeSafe AI's System One API and the Jev model. It sends state with Noul, Choice and Score questions and reads answers through typed helpers such as NoulOf, ChoiceOf and ScoreOf, with bulk accessors and a RawQuestion escape hatch for question shapes added to the API later.

Its focus is safety and robustness in services. Responses missing required fields are rejected with an error naming the field rather than read as a confident zero. The API key is masked in errors, logs and rendering, and the bearer token is not forwarded on redirects. Retry-after waits are clamped, jittered and capped, and the connection pool is sized for concurrent use. Errors map to sentinel values for rate limits, overload, authentication and timeouts, and carry the request ID. It needs Go 1.22 or newer and only the standard library, and suits Go backend teams calling Jev at volume.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this about billing?—
ChoiceWhat is the tone?angry · calm
ScoreHow urgent is this?Can wait · This week · Today

How it works

  1. Create a client with jev.New() (reads TYPESAFE_API_KEY).
  2. Call SystemOne with state and typed questions.
  3. Validate the response strictly.
  4. Branch on thresholds, for example route when billing probability is above 0.9 or send to review when confidence is low.

Strengths

  • No dependencies beyond the Go standard library.
  • Credential masking and redirect protection.
  • Bounded, jittered retries.

Caveats

  • Unofficial client.
  • Requires a TypeSafe API key.

Getting started

go get github.com/KKloudTarus/taurus-jev-sdk-go

At a glance

Type
Library
Status
beta
Language
Go
Runtime
library
Has code
yes
Has tests
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • Unofficial, dependency-free Go client for the TypeSafe AI System One API and the Jev model. Validated responses, masked credentials, bounded retries.

    hellogumbo/awesome-jev

Frequently asked questions

What makes it different from other Go clients?

It rejects incomplete responses, masks the API key everywhere and caps retry waits.

Does it need a TypeSafe key?

Yes, it reads TYPESAFE_API_KEY by default.

Can it handle new question types?

Yes, through RawQuestion and Answer.Raw without waiting for a release.

Related projects

  • ai

    Hack Club's lightweight LLM proxy, listed for adding a Jev forwarding endpoint behind its auth, limits and usage logging.

    MCP serverproductionTypeScript
  • Ax

    Ax, a DSPy-style multi-language LLM framework, listed with a native TypeSafe provider for Boolean, Choice, Score and raw Jev questions.

    PluginproductionTypeScript
  • Effect

    Effect TypeScript library whose @effect/ai-typesafe package maps classify, probability and rating operations to Jev questions.

    ProjectproductionTypeScript
  • instructor-php

    PHP monorepo whose Polyglot module adds a TypeSafe driver for typed Noul, Choice and Score decisions alongside LLM inference.

    ProjectproductionPHP
  • jev-mcp (freepik-company)

    Go MCP server giving any client typed Jev decisions via OpenRouter or TypeSafe, with validated answers and no paid retries.

    MCP serverproductionGo
  • llm-typesafe

    Simon Willison's LLM plugin that adds Jev as a model for yes/no, choice and score questions from the command line.

    PluginproductionPython