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

Jev Recipes

248 small installable TypeScript recipes that each make one typed Jev decision: route, rerank, gate, grade, compare or label.

Libraryproduction
View on GitHub ↗Live demo ↗

Overview

jev-recipes is an npm package of focused TypeScript recipes, each a typed question with its options, for agents, RAG and support flows. A recipe accepts your data, calls Jev through TypeSafe's System One API or an injected compatible client, and returns a structured decision; your application decides what happens next. The README cites 248 recipes for routing messages, checking evidence, grading and labelling model responses, comparing and gating.

Recipes can be called directly or plugged into the Vercel AI SDK or LangChain via adapters, and guidance exists for coding assistants. A recipe website hosts guides and a catalog, some recipes can be tried without a key, and a live music app demonstrates use. Versioned documentation matches release tags, and issue templates include a way to report unexpected model decisions. CI runs tests.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich route should this message take?(inferred)—
Yes / noIs this answer supported by the evidence?(inferred)yes · no
ScoreHow good is this model response on the rubric?(inferred)—

How it works

  1. npm install jev-recipes
  2. Pick a recipe for your task
  3. Call it with your data (or via an AI SDK or LangChain adapter)
  4. Use the structured decision in your application

Strengths

  • Large catalog of ready-made questions
  • Framework adapters for AI SDK and LangChain
  • Versioned docs and CI

Caveats

  • Requires a TypeSafe key or compatible client
  • Recipes still need threshold choices in your code

Getting started

npm install jev-recipes, then see the Quickstart at jev-recipes.com

At a glance

Type
Library
Status
production
Language
TypeScript
Runtime
Node library
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

  • 248 small Jev decisions for JavaScript and TypeScript: route, rerank, gate…

    wh000wh000/awesome-jev-live

  • 80+ composable TypeScript recipes powered by Jev for AI agents, retrieval, answer verification, and conversation workflows.

    hellogumbo/awesome-jev

  • Two hundred small decisions as installable TypeScript, each a typed question with its options, for routing, grading, gating, comparing and labelling.

    MrJev/awesome-jev

Frequently asked questions

What is a recipe?

A function wrapping one typed Jev question with its options for a specific task.

Does it work with LangChain?

Yes. Adapters are provided for the Vercel AI SDK and LangChain.

How many recipes are there?

The README cites 248.

Related projects

  • ruby_llm-providers-typesafe

    RubyLLM provider gem for TypeSafe Jev: block-style typed evaluations with noul, choice and score, plus RubyLLM.rerank support.

    ProjectbetaRuby
  • spring-ai-typesafe

    Java SDK for TypeSafe's System One API plus Spring AI integrations using Jev as judge, guardrail, RAG filter/reranker and tool index.

    LibrarybetaJava
  • 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
  • req\_llm

    Elixir library for unified LLM calls across providers, with an evaluate interface that calls Jev via TypeSafe or OpenRouter.

    ProjectproductionElixir