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

claude-x-jev

Free Claude Code skill where Jev sorts and Claude writes: classify, pre-route items so Claude reads only unsure ones, and gate tool calls.

ProjectbetaMIT
View on GitHub ↗

Overview

Claude x Jev is a free Claude Code skill built on the idea that Jev sorts and Claude writes. Jev makes one decision at a time in about a third of a second for a fraction of a cent and says how sure it is, so Claude only reads what needs real reasoning. It installs with npx and adds commands: /jev-setup (one OpenRouter key and a test call), /jev-classify to sort anything, /jev-route so Jev pre-sorts and Claude reads only the unsure ones, and /jev-gate for a quick safety check before tool calls.

It needs only python3 and an OpenRouter key. The README compares Jev 1.13 with Claude on an inbox-sorting task: when Jev was at least 70% sure (82% of the inbox) it matched Claude 95.5% of the time, versus 47% when less sure, and it notes cases where Claude saw deals that were not there. The author promotes a community and YouTube channel.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich category does this item belong to? (classify)—
Yes / noIs this tool call safe to run? (gate)yes · no

How it works

  1. Run /jev-setup with an OpenRouter key
  2. Classify or route a batch with Jev
  3. Claude reads only items Jev is unsure about
  4. Optionally gate tool calls

Strengths

  • Claude spends attention only on unsure items
  • Minimal requirements
  • Confidence split measured on a real inbox

Caveats

  • Comparison is the author's own single task
  • Promotional framing

Getting started

npx claude-x-jev install --with-commands, restart Claude Code, then /jev-setup

At a glance

Type
Project
Status
beta
Language
Python
Runtime
Claude Code skill
Has code
yes
License
MIT

Use cases and integrations

How curated lists describe it

Frequently asked questions

What do I need?

python3 and an OpenRouter API key.

How reliable is Jev when confident?

In the author's inbox test, it matched Claude 95.5% of the time when at least 70% sure.

What does /jev-route do?

Jev pre-sorts items and Claude reads only the unsure ones.

Related projects

  • wiring-jev-into-a-local-agent

    Educational email agent where Jev picks each message's label, reply need and family flag before drafting.

    ProjectprototypePython
  • Bops

    Mac app with AI bots running business ops on cloud computers, using TypeSafe for small yes/no calls like risk and which bot takes a job.

    Projectproduction
  • azdaja

    Recursive language model layer for coding agents that keeps inputs in a local evaluator, with optional Jev judge_many for typed calls.

    ProjectbetaRust
  • claude-jev

    Claude Code plugin where Jev decides and Claude writes: sort piles of text, calibrate thresholds, route prompts and gate Bash commands.

    Pluginbeta
  • Footwork

    Verified computer-use agent: a Jev guard checks claimed completions and gates irreversible actions; Jev also takes simple steps.

    ProjectbetaPython
  • jev-auto-approve

    Claude Code PreToolUse hook that auto-approves shell commands Jev is confident are read-only, and otherwise shows the normal prompt.

    HookbetaPython