skills.rest
Ricardo-Vae
Ricardo-VaeCOMMUNITY
Explore all skills by this author →
GitHub

agent-browser

Drive Chromium sessions to automate website interactions via accessibility-tree refs.

Updated May 13, 2026
Search Tags:#authentication#screenshots#browser automation#snapshots#chromium#ui navigation#web extraction
One-click install
npx skills add https://github.com/Ricardo-Vae/codex-research-skills --skill agent-browser-ricardo-vae
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/Ricardo-Vae/codex-research-skills/tree/main/skills/agent-browser
Command: npx skills add https://github.com/Ricardo-Vae/codex-research-skills --skill agent-browser-ricardo-vae

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This skill eliminates the friction of manually navigating websites by enabling reliable browser automation and structured page understanding for AI agents.

Core Features & Use Cases

  • Guided browser control for websites: Open pages, navigate tabs, click, fill forms, upload files, and capture screenshots/PDFs.
  • Ref-based interaction model: Use accessibility-tree snapshots that assign stable element refs for compact, agent-friendly actions, with automatic stale-ref guidance (re-snapshot after page changes).
  • Workflow support for real-world complexity: Authentication/session persistence, iframe interaction, waits tuned for SPA and dynamic UIs, and optional React/Web Vitals introspection.

Quick Start

Tell the user to: Use agent-browser to open the target URL, take an interactive snapshot, and then click or fill the form fields using the returned @eN refs.

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 extraction and UI navigation with Chromium?▼

Automate web extraction and UI navigation by driving a real Chromium session through an agent-browser CLI, using accessibility-tree snapshot refs to interact with page elements like clicking, filling forms, and extracting structured data.

What is a ref-based interaction model for browser automation?▼

A ref-based interaction model uses accessibility-tree snapshots to assign stable element refs for compact actions, providing automatic stale-ref guidance that instructs the agent to re-snapshot after page navigations or dynamic UI changes.

How do I handle authentication and session persistence in browser automation?▼

Handle authentication and session persistence in browser automation securely by utilizing state files or profile mechanisms, allowing the Chromium session to maintain login states across multiple automated interactions.

Can I interact with iframes and dynamic SPA UIs during web scraping?▼

Yes, you can interact with iframes and dynamic SPA UIs during web scraping by using a Chromium automation tool that supports semantic locators, CSS fallbacks, and waits specifically tuned for dynamic content loading.

How do I capture screenshots and PDFs for evidence in web automation?▼

Capture screenshots and PDFs for evidence in web automation by instructing the browser session to render the target page state, generating files for downstream processing or compliance verification.

Why do my element refs become stale after navigating web pages?▼

Element refs become stale after navigating web pages because the accessibility tree updates upon DOM changes, requiring a re-snapshot of the page to generate fresh, deterministic refs before attempting further UI interactions.

Related Skills

Explore More Education & Research Skills→
daltonkyemillerdaltonkyemiller
2

agent-browser

Automates web interactions via command-line interface for AI agents.

Community
Advanced
lucking7lucking7

ce-riffrec-feedback-analysis

Analyze Riffrec product feedback recordings into structured analysis reports.

Community
Advanced
teamniteoteamniteo
5

agent-browser

Automate browser navigation, form submission, and data extraction via accessibility-tree parsing.

Official
Advanced
spideynolovespideynolove
2

lightpanda

Fetch HTML content from web pages using a lightweight headless browser.

Community
Basic