codaaiteam

Jev Agent Tools — AI skill for Claude Code

AI community

Open-source single-file tools for AI agent builders, powered by Jev (TypeSafe AI's decision model): a loop detector that catches runaway agents and a task→tool router.

How to install Jev Agent Tools

This entry records only its repository, not the path inside it, so there is no exact command to give. Open codaaiteam/jev-agent-tools and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Jev Agent Tools does

Open-source single-file tools for AI agent builders, powered by Jev (TypeSafe AI's decision model): a loop detector that catches runaway agents and a task→tool router. No build. Use free: jevtypesafeai.com

Alternatives in AI

Codeburn — Free, local tool to track AI coding token usage and cost across 37 tools and agents (Claude Code, Cursor, Code 9.7k ★
Agentrq — AgentRQ: Human-in-loop realtime conversational task manager for AI Agents 1.1k ★
SoloDawn — Catches and repairs AI hallucinations before delivery: three quality gates, 31 built-in rules, a self-healing 312 ★

README

Jev Agent Tools 🧰

Small, open-source, single-file tools for people building AI **agents** — powered by **[Jev](https://jevtypesafeai.com)**, TypeSafe AI's *System One* decision model. You give Jev a situation and typed questions; it returns a typed, calibrated decision (a `choice`, `score`, or yes/no probability) in ~100–300 ms — no chat, no parsing, no hallucinations.

Each tool is a **single dependency-free `index.html`**. Open it, paste a key, use it on your own data.

Tool	What it does	Use live
🔁 Agent Loop Detector	Paste an agent's run (one step per line) → Jev grades each step progress / repeating / stuck / escalate and flags when it's looping. Copy the flagged steps.	use it →
🧭 Skill Router	Paste tasks → Jev routes each to the one tool an agent should call (web_search / run_code / sql_query / send_email / …) with confidence. Copy the routing as JSON.	use it →

Use it now

Zero setup: click a "use it live" link — runs on jevtypesafeai.com, no key.
Locally: open the tool's index.html, click ⚙︎ Jev API key, paste a key, run it on your own logs/traces/tasks. No server, no npm install.

Get a key

jevtypesafeai.com — a key in a minute, no waitlist (endpoint is browser-ready).
Or an official TypeSafe key — switch the endpoint in settings (the official API may block direct browser calls via CORS; use jevtypesafeai.com or a proxy).

Your key lives only in your browser's `localStorage`.

Why decisions, not prose

An LLM returns text you have to parse and can hallucinate. Jev returns a **typed value with a calibrated probability** you can threshold — so an agent can act on a number (`if risk < 0.2`) instead of re-reading a paragraph.