User repository on GitHub
Browser QA where Playwright drives and films, and TypeSafe Jev judges. JSON-flow CLI for agents, run dashboard, agent skill.
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Jev Browser Qa
Browser QA where Playwright drives and films, and Jev judges: natural-language assertions, self-healing locators and goal-driven click loops, with a JSON-flow CLI for agents, a local dashboard and an agent skill.
- Stars
- 4
- Language
- TypeScript
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- jonymusky
- Last verified
- Oct 2026
Used inJev for Software & DevTools
- Languages
- TypeScript
- Platforms
- PlaywrightCLIVercel AI Gateway
- Capabilities
- Open sourceRequires API keyRuns locally
Highlights
- Playwright navigates, types, waits and records; Jev answers does the claim hold, which control matches the intent, and what to click next.
- Natural-language assertions, self-healing locators and goal-driven click loops, powered by a model that answers in under 400 ms and never generates a token.
- Jev is served through the Vercel AI Gateway via the AI SDK's experimental_evaluate, so one AI_GATEWAY_API_KEY is the whole configuration.
- The CLI returns one JSON result per flow: pass/fail per step, the probability behind each judgment, a video, a final screenshot, exit code 0/1.
- Runs are recorded as MP4 when ffmpeg is installed, WebM otherwise, with a trace on failure; text entry is never delegated to the model.
Quickstart
bash
git clone https://github.com/jonymusky/jev-browser-qa && cd jev-browser-qa
pnpm install
pnpm exec playwright install chromium
cp .env.example .env
pnpm jev:smokeWatch out
Copy reviewed against the repo README on 2026-10-15; the captured README was truncated at the CLI output example, so the Numbers section was not visible and no benchmark figures are quoted here. The dashboard's nothing-leaves-your-machine claim covers the dashboard; Jev calls go through the Vercel AI Gateway with zero-data-retention routing requested per call.
Questions about this entry
Short answers drawn from the linked project or page.
Most of an end-to-end test is mechanical; the hard part is judgment — did that work, which button is create, how do I get to settings. Brittle selectors and expect(text) checks or a frontier LLM spending a full turn per click both handle it badly. Jev answers typed questions in under 400 ms and never generates a token.
Playwright handles navigation, typing, waiting, video, trace and screenshots. Jev handles the judgment: expect for does this claim hold on the page, click and pick for which visible control matches an intent, and goal or runGoal for the next click toward a goal. Text entry is never delegated to the model.
Jev sees text — URL, title, visible text and the aria tree — not pixels. That is why it can answer calibrated probability questions about a page without screenshots, and the README contrasts this with spending a full LLM turn per click.
Node 22+, pnpm, an AI Gateway key, and an app to test. ffmpeg on the PATH is optional but recommended so recordings play in Safari. Then clone, install, run pnpm exec playwright install chromium, copy .env.example to .env with your keys, and run pnpm jev:smoke to verify the model.
Run a flow and get one JSON result: pass/fail per step, the probability behind each judgment, a video and a final screenshot, with exit code 0/1. Secrets are referenced as placeholders like ${QA_EMAIL} and are never seen by the agent.
Yes. The repo ships skills/jev-qa/SKILL.md, which teaches Claude Code, Codex, Cursor and friends how to write flows and read results. The three primitives expectPage, pick and runGoal are also exported for your own Playwright tests.
Run history, KPIs, every Jev decision with probability, latency and tokens, and the video inline, in English, Spanish and Portuguese. The README says the dashboard is local and nothing leaves your machine; Jev calls themselves go through the Vercel AI Gateway with zero-data-retention routing requested per call.
More Jev resources like this
Write end-to-end tests for web apps in plain English; jev-e2e turns each case into a plan, has Jev pick the page controls, and runs it with Playwright, returning PASS, FAIL or BLOCKED with HTML, JSON and masked screenshot evidence.
Jev Tools & CLIs#community#catalog-import#madewithjev
Community
perixtar/jev-e2eTypeScript
perixtar/jev-e2eTypeScriptLocal-first test selector that asks a judge — classifier.dev by default, or Jev with a TypeSafe key — whether each test could be affected by the current code change, then hands only the selected files to Playwright or Vitest. Uncertainty always runs the full suite.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
baronunread/leanestTypeScript
baronunread/leanestTypeScriptA Playwright-based end-to-end test runner for web apps where Jev powers plain-English assertions, actions by description, goal-driven runs and failure triage. Its API mirrors Playwright Test, with test, expect, fixtures and page objects, and every Jev feature is optional.
Jev Tools & CLIs#community#github#testing
Community
neeraj719714/intelliwrightTypeScript
neeraj719714/intelliwrightTypeScriptA Node CLI for goal-driven web E2E testing: Jev picks the next step from a bounded, code-generated action space, playwright-cli executes it in a real browser, and your test code owns the deterministic assertions, thresholds and final verdict.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
dingw530/playwright-jevJavaScript
dingw530/playwright-jevJavaScriptAn enterprise admin-console test tool that generates replayable Excel test cases from live browser pages, or replays existing Excel cases, driving Playwright with local Laya or a compatible decision API to match fields, options, and targets.
Jev Tools & CLIs#community#catalog-import#madewithjev
Community
anthonyli/laya-pilotJavaScript
anthonyli/laya-pilotJavaScriptOpen-source, pre-alpha CLI that runs plain-English browser tests on Playwright, using Jev for bounded decisions: resolving each step, picking elements among offered candidates and judging verify claims with probabilities. Bring your own TypeSafe key; MIT licensed.
Jev Tools & CLIs#community#e2e-testing#playwright
Community
sedum-dev/sedumTypeScript
sedum-dev/sedumTypeScriptFrom 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.