# Jev drives a browser

> A browser automation skill that lets AI make interaction decisions step by step.

- Website: https://jev-browser.rovira.sh
- Category: AI agent builders & automation
- Launched on RankRoot: 2026-10-21
- Visitors sent by RankRoot: 0
- Show HN discussion: https://news.ycombinator.com/item?id=50181149

## About

Jev drives a browser by reading a list of interactable page elements and choosing the next action based on confidence. Each step is an independent, stateless request where the skill supplies text strings while Jev selects targets. The system validates all answers against the generated element table before execution.

## Who it's for

Developers building browser automation and agent workflows.

## Use cases

- Automating Hacker News login and search workflows
- Automating Amazon product search and cart management
- Executing multi-step browser tasks without processing DOMs or screenshots

## Features

- Reads pages and lists interactable elements for typing, clicking, or selecting
- Validates every answer against the generated element table
- Operates statelessly with independent request steps
- Handles up to 255 options per question

## FAQ

### What does Jev do?

Jev drives a browser by reading a page's interactable elements and making decisions on what to click, type, or select until a goal is complete.

### How does Jev interact with web pages?

Jev reads a table of elements built by the skill rather than viewing the DOM or screenshots, and picks which string goes where based on confidence levels.

### Is Jev stateful?

No, each step is an independent request where the only memory is the recent actions sent by the skill.


Source: https://getrankroot.com/launches/jev-drives-a-browser. Product facts come from the maker or the product's own website.