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

optim-jev

Claude Code plugin with three Jev tools: verbatim context compaction, per-edit model routing and per-prompt skill selection.

Projectbeta
View on GitHub ↗

Overview

optim-jev is a Claude Code plugin that aims to cut token spend in long sessions without changing how you work. It bundles three independent Jev-powered tools. jev-compacter, on by default, removes stale tool calls and outputs during /compact and auto-compaction instead of summarising the whole conversation. jev-router, off by default, routes file changes to the right model using project-specific rules that update when you tell Claude a preference such as letting Haiku edit tests. jev-skills, off by default, sends a short relevant skill list for each prompt instead of the full catalogue.

Each tool fails open if unavailable and has a stats command reporting session-level cost and savings in US dollars, plus why commands to explain decisions. It requires Claude Code with plugin support, Python 3.10+ and a TypeSafe API key set in /config or the environment.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this tool call or output still needed?(inferred)yes · no
ChoiceWhich model should make this file change?(inferred)—
ChoiceWhich skills are relevant to this prompt?(inferred)—

How it works

  1. Install the plugin from the local marketplace.
  2. Compaction prunes stale tool output automatically.
  3. Optionally enable routing and skill selection.
  4. Check /jev-compact stats or /jev-route why for savings and reasons.

Strengths

  • Three independent, fail-open tools.
  • Dollar-denominated savings reporting.
  • Project rules evolve with your preferences.

Caveats

  • Requires a TypeSafe key.
  • Router and skills are off by default and need setup.

Getting started

git clone https://github.com/Oqura-ai/optim-jev.git && claude plugin marketplace add ./optim-jev && claude plugin install optim-jev@optim-jev

At a glance

Type
Project
Status
beta
Language
Python
Runtime
Claude Code plugin
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

What does jev-compacter do?

Removes stale tool calls and outputs instead of summarising the conversation.

Are all tools on by default?

Only jev-compacter; jev-router and jev-skills are off by default.

How do I see savings?

Each tool has a stats command reporting cost and savings in dollars.

Related projects

  • hermes-jev-skills

    Skill pack handing an agent's small decisions to Jev: model routing, skill choice, search, memory, compaction and computer use.

    Agent skillbetaPython
  • Jev-AI-Skill

    One skill and MCP server for Claude Code, Codex and Hermes where Jev gates large-model turns, routes models and guards Git steps.

    MCP serverbetaPython
  • jev-mod

    Claude Code mod that hands small decisions (routing, skill choice, screening, gating, trimming) to Jev or another decision backend.

    ProjectbetaTypeScript
  • jev-skills (n23eos)

    Fourteen opt-in Claude Code and Codex skills where Jev makes one bounded choice: which skill, model, test, file or next step.

    ProjectbetaPython
  • JevScout

    Toolkit where Jev condenses large MCP responses for Claude Code and maps Codex requests to reviewed, fixed repository operations.

    ProjectbetaTypeScript
  • JCR (Jev Capability Resolver)

    Gives an agent one tool that uses Jev to search a nested capability tree and return only the documented commands a task needs.

    ProjectprototypeTypeScript