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

jev-awesome-skills

Open-source Jev skills for Claude Code, Codex, Cursor and Grok: one client to gate, route, guard, verify, triage, review and act.

ProjectbetaMIT
View on GitHub ↗

Overview

jev-awesome-skills, published as @aitofy/jev-awesome-skills, is a set of open-source Jev skills plus a single client that agents run instead of hand-writing a Jev script for each job. The core judgment decides when to proceed, when to ask the user and when to stop, and the same mechanism routes a skill, guards a command, verifies a claim, triages a record, reviews a diff or picks the next legal action. Skill folders include jev-gate, jev-route, jev-guard, jev-verify, jev-triage, jev-review and jev-act.

It targets Claude Code, Codex, Cursor and Grok. The package is ESM for Node 20+ with no runtime dependencies, reads the key only from TYPESAFE_API_KEY and never prints it, and supports dry-run and offline fixture modes that need no key, for example decideFixture for trying the decision logic. The README suggests a prompt telling the agent to use the bundled jev command, dry-run first, and not write a second HTTP client. It has CI and an MIT licence.

What the decision model decides

TypeQuestionAnswers
ChoiceProceed, ask the user, or stop?(inferred)proceed · ask · stop
Yes / noIs the agent determined/confident enough to act?yes · no
ScoreHow harmful is this action, from reversible to destructive?—

How it works

  1. Install the package and copy skill folders into the agent's skills directory
  2. Agent builds a request for gate, route, guard, verify, triage, review or act
  3. Dry-run, then call Jev via the bundled client
  4. Decision logic returns proceed, ask or stop

Strengths

  • One client for many skill types
  • Offline fixtures for testing without a key
  • No runtime dependencies

Caveats

  • Requires a TypeSafe key for live use
  • Skill quality depends on agent following instructions

Getting started

`npm i @aitofy/jev-awesome-skills`, copy the skills/ folders into ~/.claude/skills/, then `npx jev --dry-run --file request.json`.

At a glance

Type
Project
Status
beta
Language
TypeScript
Runtime
Node CLI, agent skills
Has code
yes
Has tests
yes
License
MIT

Use cases and integrations

How curated lists describe it

Frequently asked questions

What skills are included?

Gate, route, guard, verify, triage, review and act, sharing one Jev client.

Can I try it without a key?

Yes, dry-run and fixture modes need no key.

Which agents are supported?

Claude Code, Codex, Cursor and Grok.

Related projects

  • 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
  • jev-classifier

    Local gateway and MCP tools that let Jev predict which tool a coding agent should call next, from suggestions to enforced choices.

    MCP serverbetaTypeScript
  • jevmem

    Project memory for Claude Code that asks Jev whether each chat statement is worth keeping and checks tool calls against saved rules.

    ProjectbetaTypeScript
  • master-jev-hook

    Local gateway, MCP server and Claude Code hooks that let coding agents delegate explicit choices, classification and scoring to Jev.

    MCP serverbetaTypeScript
  • jev-agent-kit

    Typed judgment layer for Claude Code, Codex and Hermes Agent: deterministic facts in code, language judgments in Jev, with receipts.

    ProjectbetaPython
  • 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