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

claude-jev (darwintechlab)

OpenJev for Claude Code: plugin and MCP server giving Claude jev_choice, jev_noul, jev_score and jev_ask tools on the live API.

MCP serverbetaMIT
View on GitHub ↗

Overview

OpenJev for Claude Code, from Darwin Tech Lab, gives Claude Code a model built for small decisions with few possible answers: is this command safe, which tool fits, is this PR ready, how serious is this bug. You pass text and possible answers and get back which fits and how confident it is, always one of your options.

Installing the plugin adds five tools through a bundled MCP server: jev_choice, jev_noul, jev_score, jev_ask and jev_doctor. The README cites a live benchmark with a p50 of 307 ms and about $0.017 per 1,000 decisions, and stresses that calibrated confidence lets the agent act when sure and ask when not, illustrated with a ticket that lacks enough detail getting a low score. It mirrors the opencode-openjev plugin, is live-only (no local model) and requires Node 20+.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich team should handle this ticket?—
Yes / noIs this command safe to run?yes · no
ScoreHow serious is this bug?—

How it works

  1. Claude reaches a small decision
  2. Calls jev_choice, jev_noul, jev_score or jev_ask
  3. Gets the answer with confidence
  4. Acts when confident, asks the user when not

Strengths

  • About 0.3 s per decision per the README
  • Calibrated confidence for act-or-ask
  • Doctor tool for setup checks

Caveats

  • Live API only
  • Benchmark figures are the project's own

Getting started

Follow SETUP.md (about three minutes) to install the plugin and configure your TypeSafe key.

At a glance

Type
MCP server
Status
beta
Language
TypeScript
Runtime
Claude Code plugin + MCP server (Node 20+)
Has code
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • Jev System One for Claude Code — typed Choice/Noul/Score via live TypeSafe API (mirror of opencode-openjev, live-only MCP)

    logicrw/awesome-jev-projects

  • Jev System One for Claude Code — typed Choice/Noul/Score via live TypeSafe API…

    wh000wh000/awesome-jev-live

  • Jev System One for Claude Code — typed Choice/Noul/Score via live TypeSafe API (mirror of opencode-openjev, live-only MCP).

    hellogumbo/awesome-jev

Frequently asked questions

What tools does Claude get?

jev_choice, jev_noul, jev_score, jev_ask and jev_doctor.

How fast is it?

A live benchmark p50 of 307 ms, per the README.

Is there an OpenCode version?

Yes, this mirrors opencode-openjev.

Related projects

  • typesafe-mcp (MarkChu-git)

    Bun-based MCP server wrapping Jev with compact answers and a code-computed act, review or abstain verdict.

    MCP serverbetaTypeScript
  • jev-fuse

    Governance proxy between Jev or local Laya and agents that turns probabilities into allow, ask, deny, keep, truncate or drop actions.

    MCP serverbeta
  • actiongate-jev

    Gateway that checks agent tool calls with policy plus Jev evidence, then binds approval to a single-use permit for that exact action.

    MCP serverbetaTypeScript
  • hermes-jev (DoGMaTiiC)

    Two stdlib-only Hermes Agent plugins: a Jev pre-tool gate with a jev_ask tool, and a skill router that names one skill per turn.

    Agent skillbetaPython
  • jev-agent-authorization

    Starter kit that gates every MCP tool call with Kinde identity checks plus Jev risk signals, then allows, holds for approval or stops it.

    MCP serverbetaTypeScript
  • jevguard-mcp

    Zero-dependency Python MCP server where Jev gates shell commands, audits patches and resolves decisions, with SQLite caching.

    MCP serverbetaPython