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

jev-cli

Rust agent-facing CLI built on the unofficial jev-sdk crate: predict, choose and score with strict stdout and meaningful exit codes.

CLIbetaMIT OR Apache-2.0
View on GitHub ↗

Overview

jev-cli by portlandhodl is an agent-facing command line for Jev built on the unofficial jev-sdk Rust crate. Its premise is that LLM agents act first and find out later; with jev-cli an agent asks a typed question before acting and gets calibrated probabilities to gate on or to pick among paths, in a single shell call. predict returns the probability that a statement holds (for example, will this deploy complete without errors), choose picks among labelled options (retry, fallback or abort), and score returns a probability-weighted value on a rubric (trivial, moderate, severe).

Because it is built for scripts and tool-use loops, stdout carries only the answer (JSON by default), diagnostics go to stderr, and exit codes are meaningful, with optional thresholds and pass fields. It installs with cargo from GitHub and reads the key from a permission-restricted dotenv-style config, a project .env or environment variables, in defined precedence. It is dual-licensed MIT or Apache-2.0 and not affiliated with TypeSafe.

What the decision model decides

TypeQuestionAnswers
Yes / noWill this deploy complete without errors?yes · no
ChoiceWhich approach is most likely to succeed?retry · fallback · abort
ScoreHow risky is this plan?trivial · moderate · severe

How it works

  1. Agent prepares context
  2. Calls jev-cli predict, choose or score
  3. Reads JSON from stdout and checks exit code
  4. Gates or chooses its next action

Strengths

  • Strict output contract for agents
  • Thresholds with pass/fail
  • Rust binary with no runtime

Caveats

  • Unofficial SDK and CLI
  • Requires building with cargo

Getting started

`cargo install --git https://github.com/portlandhodl/jev-cli`, then put TYPESAFE_API_KEY in ~/.jevcli/jev.conf.

At a glance

Type
CLI
Status
beta
Language
Rust
Runtime
CLI
Has code
yes
License
MIT OR Apache-2.0
Stars
2

Use cases and integrations

How curated lists describe it

  • Agent-facing command line for Jev, TypeSafe AI's System One model: ask typed questions, get calibrated probabilities. Unofficial, built on jev-sdk.

    wh000wh000/awesome-jev-live

Frequently asked questions

What commands exist?

predict, choose and score.

Why strict stdout?

So agents and scripts can parse answers without noise; diagnostics go to stderr.

Is it official?

No, it is built on the unofficial jev-sdk crate.

Related projects

  • decision-gate

    npm library that rate-limits, spend-caps, caches and redacts every Jev request made in a loop.

    LibrarybetaJavaScript
  • jev-decision-hub

    Russian-language Jev toolkit with a zero-dependency Python client, key pooling and a guide to free credits via a third-party relay.

    LibraryprototypePython
  • KISS

    Fast Rust terminal coding agent with 44 providers that can ask Jev for its own decisions through a bundled client.

    LibrarybetaRust
  • Agent Orchestration SDK

    Orchvia: local engine with TS and Python SDKs that runs Claude Code and Codex agents as a team, routing tasks by code and typed decisions.

    LibrarybetaTypeScript
  • 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