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

pi-jev-core

Minimal Pi extension and TypeScript client registering a jev_evaluate tool for noul, choice and score across five Jev platforms.

Projectbeta
View on GitHub ↗

Overview

pi-jev-core is a minimal, standalone Pi extension that connects to a Jev API and registers a jev_evaluate tool supporting noul, choice and score judgments. It deliberately contains no tool routing, auto mode, skill discovery or context compaction, and can also be imported programmatically as JevClient in TypeScript apps.

It supports five platforms: TypeSafe (default, jev-latest), OpenRouter (typesafe/jev-1.13), Cloudflare (typesafe/jev, needing account and gateway ids), Vercel AI Gateway (typesafe-ai/jev) and a local Jev API server on localhost with no key. Credentials come from environment variables or Pi secret files, and a /jev-platform command lists, switches and persists the active platform and logging. It is documented in English and Chinese.

What the decision model decides

TypeQuestionAnswers
Yes / noAny yes/no question the agent asks via jev_evaluateyes · no
ChoiceAny choice question—
ScoreAny score question—

How it works

  1. Install and choose a platform with /jev-platform.
  2. Agent calls jev_evaluate with state and questions.
  3. The client sends the request to the selected platform.
  4. Typed answers return to the agent.

Strengths

  • Five interchangeable platforms including local.
  • Small, focused codebase.
  • Usable outside Pi as a library.

Caveats

  • No higher-level features; you build the policy.
  • Ships TypeScript source loaded by jiti.

Getting started

pi install git:github.com/morler/pi-jev-core and set TYPESAFE_API_KEY (or another platform's credential).

At a glance

Type
Project
Status
beta
Language
TypeScript
Runtime
Pi extension, library
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

Which platforms are supported?

TypeSafe, OpenRouter, Cloudflare, Vercel AI Gateway and a local server.

Does it route models or compact context?

No, it only provides the jev_evaluate tool.

Can I use it outside Pi?

Yes, import JevClient from pi-jev-core.

Related projects

  • pi-typesafe-jev

    Pi extension exposing Jev as five narrow tools (noul, choice, score, evaluate, pinned ask) while code keeps thresholds and actions.

    PluginbetaTypeScript
  • 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
  • adk-go-typesafe

    Go integration bringing TypeSafe Choice, Score and Noul to Google's adk-go agents, with a typed client, routing agent and callbacks.

    ProjectbetaGo
  • feelings

    Ruby gem for probabilistic conditionals like Feelings(msg).like?("urgent"), answered by Jev or other decision models.

    ProjectbetaRuby
  • fuzzy-jev

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

    CLIbetaRust
  • Gut

    Elixir DSL (Gut.feel) that picks an Elixir value for a subject and question, with a Jev evaluation adapter among LLM options.

    ProjectbetaElixir