skills.rest
davekilleen
davekilleenCOMMUNITY
Explore all skills by this author →
GitHub

agent-browser

Automate browser tasks via CLI using accessibility snapshots and ref-based element selection.

461|129|Updated Jan 26, 2026
Search Tags:#screenshots#AI agent#browser automation#CLI#web scraping#form filling
One-click install
npx skills add https://github.com/davekilleen/Dex --skill agent-browser-davekilleen
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/davekilleen/Dex/tree/main/.claude/plugins/compound-engineering/skills/agent-browser
Command: npx skills add https://github.com/davekilleen/Dex --skill agent-browser-davekilleen

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill automates web browsing tasks, allowing AI agents to interact with websites, fill forms, take screenshots, and scrape data without manual intervention.

Core Features & Use Cases

  • Web Navigation: Open, reload, and navigate back/forward through web pages.
  • Element Interaction: Click, type, fill, select, check, and scroll elements identified by refs or semantic locators.
  • Data Scraping & Screenshots: Extract text, HTML, attributes, and capture full-page or viewport screenshots.
  • Use Case: Automatically fill out a complex online application form, navigate through search results, and save a screenshot of the confirmation page.

Quick Start

Use the agent-browser skill to open the website 'https://example.com' and then click on the element with ref 'e1'.

Frequently Asked Questions about agent-browser

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate web browser tasks using an AI agent?▼

You can automate web browser tasks by using the agent-browser Skill, which enables AI-driven browser automation through a CLI to navigate pages, fill forms, and scrape data. It relies on accessibility snapshots and ref-based element selection for deterministic execution.

Can I use AI to fill out complex online forms automatically?▼

Yes, you can use AI to fill out complex online forms automatically. The Skill supports element interaction like typing, selecting, and checking elements identified by semantic locators or refs to complete form submissions without manual intervention.

How do I take full-page screenshots during web scraping?▼

To take full-page screenshots during web scraping, the Skill supports capturing full-page or viewport screenshots. It extracts text, HTML, and attributes from web pages while performing automated navigation tasks.

Do I need Vercel's agent-browser CLI to run AI browser automation?▼

Yes, you need Vercel's agent-browser CLI and Chromium installed to run this AI browser automation. These dependencies are required to execute the command-line interface that interacts with web pages via accessibility snapshots.

How does an AI agent identify elements on a web page for automation?▼

An AI agent identifies elements on a web page for automation by using accessibility snapshots and ref-based element selection. This method targets specific elements through semantic locators to perform actions like clicking, typing, and scrolling.

What are the limitations of automating web interactions with AI?▼

Limitations of automating web interactions with AI include the requirement for Vercel's agent-browser CLI and Chromium. The automation relies on ref-based element selection and accessibility snapshots, which may require deterministic page structures to function correctly.

Related Skills

Explore More Software Engineering Skills→
marchattonmarchatton
5

agent-browser

Automate web browser interactions via CLI for navigation, form filling, and data extraction.

Community
Intermediate
bendruckerbendrucker
16

json

Process JSON data with jq filters for parsing, filtering, and transforming.

Community
Basic
jackspacejackspace
16

flowio

Parse FCS files to extract events as NumPy arrays and metadata.

Community
Advanced
isndotbizisndotbiz

agent-browser

Automate web browser interactions via the agent-browser CLI.

Community
Intermediate