CommunityJev Demos & Experiments
Putting Jev in a Playwright testing loop
A field report on putting Jev inside a Playwright browser-testing loop: two Jev calls per probe, 63 probes in 24.3 seconds, and a 131 ms median round trip, with a candid account of near coin-flip decisions.
- Published by
- Community
- Author
- Jason Arbon
- Last verified
- Oct 2026
Highlights
- Two Jev calls per probe: one chooses the next action, one judges whether the result looks wrong.
- 63 probes in 24.3 seconds of active testing, 126 calls, and a 131 ms median round trip.
- Jev sees page text and structured observations, not screenshots; Playwright and code do the rest.
- The author reports near coin-flip verdicts on some cases, and recommends assertions first.
Watch out
A single-authored field report on one buggy storefront, not an independent benchmark; the accuracy caveats are the point of the piece, so read it before copying the pattern.
Questions about this entry
Short answers drawn from the linked project or page.
It put Jev inside a Playwright browser-testing loop against a deliberately buggy storefront. The controller offered a menu of probes, Jev chose which to run next, then judged whether the result looked wrong.
Two Jev calls per probe, 63 probes in 24.3 seconds of active testing, 126 Jev calls total, and a 131 ms median round trip, with the model identified as jev-1.13.0 in the recorded responses.
The author is candid: one normal checkout scored 0.49 for expected behavior against 0.47 for a bug, a near coin flip, and an earlier run falsely flagged an ordinary add-to-bag. Accuracy was uneven.
Use deterministic assertions where the expected result is precise, and reserve Jev for the judgments a rule cannot make. The model picks the sequence; your code defines the available moves and what correct means.
No. The controller sends page text and structured observations such as product names, quantities, totals, and status messages. Jev does not watch the recording or inspect screenshots in this implementation.
Almost everything. HTML, CSS, and JavaScript provide the app under test, the Node controller builds the action menu and enforces limits, and Playwright drives the browser and reads the page.
The write-up is on the author's Medium, cross-posted to Substack, by Jason Arbon, who works on AI-driven testing at testers.ai and previously built test.ai. It is a field report, not a benchmark.
More Jev resources like this
97GitHub stars
A Playwright-based CodexQA skill where Jev chooses indexed controls on a page-built index while the browser executes the action.
Jev Demos & Experiments#community#browser#playwright
Community
openqa-cn/jev-browserTypeScript
openqa-cn/jev-browserTypeScript29.3kGitHub stars
The Cua computer-use platform documents a jev-use recipe where TypeSafe Jev picks bounded action IDs and Cua Driver executes clicks and fills in the browser.
Jev Demos & Experiments#community#browser#computer-use
Community
trycua/cuaRust
trycua/cuaRust306GitHub stars
Local macOS computer-use automation on Apple silicon using on-device laya typed decisions and CUA-S1-FORMS for filling, via the Accessibility API.
Jev Demos & Experiments#community#browser#local
Community
FluidInference/FluidUseSwift
FluidInference/FluidUseSwift221GitHub stars
An agent skill that reproduces Jev-style single-step logits over indexed browser controls using local Qwen weights for offline browser use.
Jev Demos & Experiments#community#browser#open-models
Community
APUS-AI-Lab/fast-browser-usePython
APUS-AI-Lab/fast-browser-usePythonA toy that hands Jev control of page layout: it picks the size, color, and position of on-screen components and rearranges them to its own preference, exposed as a live site.
Jev Demos & Experiments#community#ui#agents
Community
An open-source data analyst agent that keeps Qwen for generating SQL and uses Jev as a review layer at three checkpoints: question clarity, SQL fidelity to the request, and whether the final explanation is grounded in the returned rows.
Jev Demos & Experiments#community#agents#sql
Community
From the Jev guides
Original write-ups on what this entry is for.
A test loop is a long series of small judgments. What the Playwright, mobile QA, and CI experiments report: two calls per probe, 131 ms medians, and near coin-flip failures to design around.
Jev Practices & Patterns#testing#qa#confidence
Read article
A game tick is a state plus a legal action list. How Clash Royale, Minecraft, and the earlier Mario, Pokémon, StarCraft, and drone loops ask Jev one step at a time.
Jev Demos & Experiments#gaming#agents#control
Read article
From the Jev community
Posts from builders shipping with Jev right now.