jev-browser-use: Choose Browser Actions from DOM or A11y Data
You should be able to explain jev-browser-use in four sentences: what enters the system, what Jev judges, what code executes, and what independently proves the job finished.
Use caseLimit browser or computer actions to reviewable choices
Stable browser agents send observations to Jev, pick a finite next action, then let code click. Pair with practice pages or jev-for-chrome. Controllable envs only — no real exam cheating.
Source · wy-coliney/jev-browser-use · Open-source project
What this project does
Section titled “What this project does”Free-form coordinates are hard to audit. Build state from the DOM/a11y tree, Choice the next step, execute in code. Prove stop conditions on a practice page first.
Sources
Section titled “Sources”
GitHub Open Graph preview for the wy-coliney/jev-browser-use repository; a repository preview, not a screenshot of running software. Notes are based on its public source.
A useful first read
Section titled “A useful first read”Limit browser or computer actions to reviewable choices
Based on the public repository; this site has not reproduced the full project. Read the source before deciding whether to try it.