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

jev-mod

Claude Code mod that adds $.jev so other mods can ask typed Choice, Score and Noul questions, for example to deny destructive commands.

Projectbeta
View on GitHub ↗

Overview

jev is a Claude Code mod that adds a $.jev object to the engine, letting other mods ask TypeSafe's Jev typed Choice, Score and Noul questions and get probabilities back. The README's example checks whether a shell command deletes files that cannot be reproduced and denies it if the Noul probability exceeds 0.9.

It calls the TypeSafe API only, with a 4-second default timeout (9 s maximum) after which the request is abandoned. Every answer is checked against its question before return, and error messages include the HTTP status and a hint but never the response body. Nothing is sent until a mod calls $.jev.ask. Mods that declare jev as a dependency get its TypeScript types. The key can be stored in the keychain via plugin options or exported as TYPESAFE_API_KEY, the same variables used by the author's jev-judge-mcp.

What the decision model decides

TypeQuestionAnswers
Yes / noThe command deletes files that cannot be reproduced.yes · no

How it works

  1. Install the mod from the gecm0/jev-mod marketplace
  2. Another mod calls $.jev.ask with state and typed questions
  3. Jev answers within the timeout; answers are validated
  4. The calling mod acts, for example denying a command above 0.9

Strengths

  • Reusable decision primitive for any Claude Code mod
  • Answers validated before return
  • Safe error messages without response bodies
  • Offline plugin tests with a fake fetch

Caveats

  • Requires Claude Code v2.1.287 or later
  • TypeSafe API only
  • Timed-out requests are abandoned, not cancelled

Getting started

/plugin install jev --marketplace gecm0/jev-mod, then set the key in plugin options or export TYPESAFE_API_KEY.

At a glance

Type
Project
Status
beta
Language
TypeScript
Runtime
Claude Code mod
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

  • Claude Code mod: adds `$.jev` to the engine so other mods can ask typed `Choice`, `Score` and `Noul` questions and get calibrated probabilities back.

    yibie/awesome-jev

Frequently asked questions

What does jev-mod add?

A $.jev.ask function other Claude Code mods can use to ask typed Jev questions.

When is data sent?

Only when a mod calls $.jev.ask, and only to api.typesafe.ai.

Can I test it without a key?

Yes. claude plugin test . runs offline with a fake fetch.

Related projects

  • jev-mcp

    MCP server giving any agent Jev's classify, score, yes/no check and allow/confirm/block gating of risky tool calls.

    MCP serverbetaJavaScript
  • jev-use

    Claude Code, Codex and pi plugin that hands non-writing agent steps to Jev and escalates writing or unsure steps to the LLM.

    PluginbetaJavaScript
  • Jevbridge

    Zero-dependency Node library with MCP and ACP adapters adding Jev decisions and gated computer use beside any LLM agent.

    MCP serverbetaTypeScript
  • jev-mcp

    Unofficial, read-only MCP server exposing Jev as four advisory tools: decide, route, risk score and yes/no gate.

    MCP serverbetaTypeScript
  • jevcore

    Jev decision layer for DeepSeek Harness, MCP hosts and plain Node, offline by default with disclosed egress and opt-in gates.

    MCP serverbetaTypeScript
  • typesafe-jev-opencode

    Local MCP server that adds a jev_decide tool to OpenCode agents, with a conservative confidence-gated policy for a developer agent.

    ProjectbetaJavaScript