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

Helm

Orchestrator that probes which coding agents are installed and logged in, then has Jev pick the best one for a task.

Agent skillbetaMIT
View on GitHub ↗

Overview

Helm steers the coding agents already on your machine, such as Claude Code, Codex, Cursor, Gemini CLI, Aider, OpenCode, Copilot and local models. It installs as a Claude Code plugin, a Gemini CLI extension or an Agent Skill; you ask your agent which agent should do a task, and Helm answers with the best installed agent, the exact command to run, and the reason.

Its design is that the probe observes, Jev judges and the chosen agent generates. A probe builds the answer set from agents that are actually installed and logged in, so an uninstalled agent cannot be recommended, and Jev picks among them. The README frames other judgment calls in orchestration, such as how risky a task is and whether an agent that exited successfully actually finished, as too costly to spend a frontier model on for every task.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich installed coding agent should take this task?—

How it works

  1. Probe installed and logged-in agents
  2. Build the answer set from what is available
  3. Jev chooses the best agent for the task
  4. Return the agent, command and reason

Strengths

  • Cannot recommend uninstalled agents
  • Works across many coding agents
  • Installs in several hosts

Caveats

  • Requires a TypeSafe key
  • Recommendation only; you run the command

Getting started

/plugin marketplace add Jimuelle07/Helm, /plugin install helm@helm, then python skills/helm/scripts/route.py --set-api-key <key>.

At a glance

Type
Agent skill
Status
beta
Language
Python
Runtime
Claude Code plugin, Gemini CLI extension, Agent Skill
Has code
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • Picks which of the coding agents on your machine should take a task, from an answer set a probe builds, so an agent you have not installed cannot be recommended. Installs as a Claude Code plugin, a Gemini CLI extension, or an Agent Skill.

    MrJev/awesome-jev

Frequently asked questions

What does Jev decide?

Which installed coding agent should handle a task.

Can it pick an agent I do not have?

No, the answer set comes from a probe of installed agents.

Where does it install?

As a Claude Code plugin, Gemini CLI extension or Agent Skill.

Related projects

  • jev-agent-hooks

    Hooks for Claude Code, Codex and pi that use Jev to suggest a skill each turn, plus a rule gate for subagent model routing.

    HookbetaJavaScript
  • jev-for-all

    Experimental plugin wiring Jev into coding-agent harnesses to pick skills, tools, models and browser moves, starting with OpenCode.

    Agent skillprototypeTypeScript
  • loadout

    Chooses the model tier, skills and MCP servers for each coding-agent session with two Jev requests, then launches Claude Code or Codex.

    MCP serverbetaPython
  • jev-pilot

    Claude Code plugin that asks Jev before each turn for reasoning effort, subagent model, strategy and skill, with a decision log.

    Agent skillbetaTypeScript
  • jev-harness-router

    Per-turn router: one ~350 ms Jev call picks model tier, effort, tools and skill, with a hard deadline and regex fallback.

    Agent skillbetaTypeScript
  • jev-decision-layer

    MIT plugin and local MCP server that hands an agent's closed questions to Jev or local Laya, with policy gates and receipts.

    Agent skillproductionPython