Skip to content

Jev Studio

pip-installable Jev kit: a CLI for verify, screen, classify, extract, route and rerank, plus an MCP server for the primitives.

MCP serverbeta

Overview

Jev Studio is a one-stop kit for experimenting with TypeSafe's Jev. A single pip install provides two commands: `jev`, a CLI that wraps common judgments over piped text, and `jev-studio`, an MCP server that serves Choice, Noul and Score tools over stdio along with prompt libraries and slash commands for TypeSafe cookbook recipes.

The CLI covers verifying claims against evidence, screening text for prompt injection or boilerplate with pass/review/block/skip recommendations, single, multi-label and hierarchical classification with confidence-gated auto or review, regex-proposed value extraction where Jev picks, record matching (same, different, unclear), routing a request to a handler with closed-set arguments, ranking up to 250 candidates, reranking, transcript compaction and batch runs. Output can be JSON, Markdown, JSONL, CSV or tables, and a --fail-on flag makes it usable as a CI gate. Authentication accepts a TypeSafe key, OpenRouter key or Cloudflare credentials, with keychain storage.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this claim supported by the evidence?(inferred)—
ChoiceAre these two records the same, different or unclear?same · different · unclear
ChoiceShould this text pass, be reviewed, blocked or skipped before an agent reads it?pass · review · block · skip

How it works

  1. Pipe text or JSONL into a jev subcommand (or call via MCP)
  2. Jev returns typed answers with probabilities
  3. Confidence gates choose auto versus review
  4. --fail-on exits non-zero for CI when a condition trips

Strengths

  • Many common decision patterns in one install
  • Multiple output formats and CI-friendly exit codes
  • Works with TypeSafe, OpenRouter or Cloudflare credentials
  • Published on PyPI

Caveats

  • Broad tool surface; quality of each command not benchmarked in the README
  • Requires an API key from a supported provider

Getting started

pip install jev-studio, then `jev auth login`.

At a glance

Type
MCP server
Status
beta
Language
Python
Runtime
CLI / MCP server
Has code
yes

Use cases and integrations

How curated lists describe it

  • If you're experimenting with jev it will be easier from here.

    hellogumbo/awesome-jev
  • One pip install for experimenting: MCP tools for Choice, Noul and Score, prompt libraries and a slash command per cookbook recipe.

    MrJev/awesome-jev

Frequently asked questions

What commands does the jev CLI offer?

verify, screen, classify, extract, match, route, ask, find, rerank, compact and batch, plus auth and config.

Can I use it in CI?

Yes, --fail-on makes the process exit 2 when a chosen judgment condition trips.

Which keys does it accept?

A TypeSafe key, an OpenRouter key, or Cloudflare API token and account ID.

Related projects

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

    MCP serverproductionGo
  • Unofficial hosted or local Jev MCP server giving agents a quick second opinion: it picks yes/no, scale or choice and returns probabilities.

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

    MCP serverbetaRust
  • Agent Client Protocol agent that exposes Jev Choice, Score and Noul decisions in IDEs with guided input and probability displays.

    ProjectbetaTypeScript
  • Small dependency-free CLI and stdio MCP server for Jev's noul, choice and score judgments over text or JSON.

    MCP serverbetaPython
  • Dependency-free CLI and Codex skill that send local state to Jev for typed judgments and return answers without echoing the state.

    Agent skillbetaPython