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

jev (drpaneas)

Unofficial standard-library Go client for Jev with generic typed Choice values and local confidence thresholds.

ProjectbetaMIT
View on GitHub ↗

Overview

jev by drpaneas is an unofficial Go package for typed AI decisions with Jev. Constructors build reusable questions locally: Choice over your own typed Go values (for example a Team string type), Noul for P(yes) and Score for a probability-weighted rubric position. Ask calls Jev, and Resolve applies your confidence threshold locally, returning a typed value or a flag that human review is needed.

It uses only the standard library. Requests have a 30-second total timeout and up to two retries, which the README warns may each be billed; retries can be disabled. Docs cover batching, configuration and errors, with a complete triage example and an offline example test. The README advises choosing thresholds from your own evaluation data.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich team should handle this ticket?billing · support

How it works

  1. go get the package and set TYPESAFE_API_KEY
  2. Build Choice, Noul or Score questions with typed options
  3. Call Ask with the state
  4. Resolve with a threshold; route low confidence to review

Strengths

  • Generic typed results in Go
  • Standard library only
  • Explicit retry and timeout controls

Caveats

  • Unofficial client
  • Retries may be billed

Getting started

`go get github.com/drpaneas/jev`, set TYPESAFE_API_KEY, then call `client.Ask(ctx, state, question)` and `decision.Resolve(0.90)`.

At a glance

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

Use cases and integrations

How curated lists describe it

Frequently asked questions

What makes it Go-friendly?

Choice options are your own typed Go values, returned as Decision[T].

Where are thresholds applied?

Locally in Resolve, so you can tune them on your own data.

Does it retry?

Up to two retries by default, which may be billed; set WithMaxRetries(0) to disable.

Related projects

  • fuzzy-jev

    Rust CLI and library that asks Jev typed questions via OpenRouter and combines the probabilities with fuzzy-logic rules into decisions.

    CLIbetaRust
  • jev (shaharia-lab)

    Rust CLI for Jev: ask yes/no, choice and rubric questions about text and branch on exit codes, JSON or MCP tools.

    MCP serverbetaRust
  • questions

    TypeScript library for typed semantic decisions via Zod schemas or native batches, defaulting to Jev, with confidence-gated policies.

    LibrarybetaTypeScript
  • Tiltmeter

    Monitoring proxy and client that records Jev answer probabilities and alerts on version changes, drift, edge crowding or accuracy drops.

    ProxybetaPython
  • daf-jev

    Composable Python toolkit for Jev: question builders, confidence gates, batch evaluator, calibration, CLI, MCP server and agent skill.

    MCP serverbetaPython
  • discern

    Effect library turning Jev Choice, Noul and Score answers into typed branches with a compiler-enforced Uncertain case.

    LibrarybetaTypeScript