Skip to content
JevDirectory.orgTraffic count unavailable. View site traffic

Jev directory

CommunityJev Tools & CLIs

Jevaluate

Maykana's free tool that lets a cheap AI test a web app like a person: write checks in plain sentences, Jev picks the next step from a short list, and below 80% confidence it stops and hands over to a human instead of clicking. A DeepSeek vision pass reports visual defects.
See highlightsOpen github.com
Stars
1
Language
JavaScript
License
MIT
Reading time
3 min
Published by
Community
Author
elshinq
Last verified
Oct 2026
Tagscommunitycatalog-importlogicrwqa-automationplaywrightevals
Works forJev for EvaluationJev for Decision gating
Used inJev for Software & DevTools
Languages
JavaScriptPython
Platforms
PlaywrightCLIBrowser
Capabilities
Open sourceRequires API key

Highlights

  • Jev only picks the next step from a short list and reports how sure it is; under 80% confidence the walk refuses to click and leaves a note for a person.
  • On the team's demo app, with 3 problems hidden on purpose, it found all 3, and the whole run cost less than one cent.
  • In a 60-request trust test in English, Arabic and mixed Arabic-English it got 10 wrong — and was unsure on all 10; whenever it was sure, it was right.
  • Each page gets one screenshot checked by DeepSeek for visual problems: text covering a button, text too faint to read, leftover code on screen.
  • Beyond the walk: a Jev accuracy eval, an on-screen text checker, a first-click menu tree test, and a skill file for Claude Code and other agents.

Quickstart

bash
git clone https://github.com/ElshinQ/jevaluate && cd jevaluate
npm init -y && npm i -D @playwright/test@1 && npx playwright install chromium
export TYPESAFE_API_KEY=...  DEEPSEEK_API_KEY=...
node scripts/walk.mjs --spec examples/walk-watch.json --watch

Watch out

Grounded on the truncated README and GitHub metadata read on 2026-10-15; the capture ends mid expected-output line from the five-minute start, so only commands shown above the cut are quoted. The trust test is 60 requests, which the README itself calls small, and screenshots go to DeepSeek — the project advises a test copy of your app and throwaway logins.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Microsoft-Decision-1 against Jev on a private eval

X: Vals AI's independent evaluation of Jev

X: An eval suite for System One models

X: Jev and Laya play chess instead of benchmarks

From the Jev guides

Original write-ups on what this entry is for.

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai