Verified:GitHub

agent-browser

Automate browser interactions for web UI testing with screenshots and JSON outputs.

Updated May 13, 2026

Search Tags:#data extraction #browser automation #web testing #form filling #screenshot capture #UI interaction #E2E workflows

Ready to run

npx skills add https://github.com/cliffthelin/AXiomEngine --skill agent-browser-cliffthelin

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/cliffthelin/AXiomEngine/tree/main/archon/.claude/skills/agent-browser
Command: npx skills add https://github.com/cliffthelin/AXiomEngine --skill agent-browser-cliffthelin

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser (npm install -g).

What problem does it solve?

It removes the need to manually drive a browser to test web apps, fill forms, and collect screenshots or extracted information.

Core Features & Use Cases

Web interaction automation: Navigate pages, click elements, fill inputs, and perform common UI workflows using refs or semantic locators.
Evidence and extraction: Capture screenshots/PDFs and extract page text, HTML, attributes, values, and page metadata.
Test orchestration support: Wait for specific UI states, scope searches, and record sessions for later review (including parallel sessions).

Quick Start

Use the agent-browser skill to open a URL, snapshot the page for element refs, then click and fill fields to submit a form and verify the result with a follow-up snapshot.

Frequently Asked Questions about agent-browser

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

FAQPage Schema

How do I automate browser interactions for end-to-end web testing?▼

Browser interaction automation for end-to-end web testing is achieved by navigating pages, filling inputs, and clicking UI elements using refs from interactive snapshots. You can wait for specific UI states, capture screenshots, and extract page data to verify form submissions and results.

Can I extract page data and capture screenshots during automated web navigation?▼

Yes, extracting page data and capturing screenshots during automated web navigation is fully supported. The process outputs machine-readable JSON alongside captured screenshots and PDFs, allowing you to collect text, HTML, attributes, and page metadata across different navigation states.

Do I need to install external tools to automate form filling and UI workflows?▼

Yes, you need to install the external agent-browser npm package globally to automate form filling and UI workflows. Once installed, you can open a URL, snapshot the page for element refs, and perform deterministic actions like filling fields and submitting forms.

What is the best way to handle authentication flows and form submission in browser automation?▼

The best way to handle authentication flows and form submission in browser automation is to use semantic locators and element refs from page snapshots. This method supports waiting for specific UI states and recording sessions for later review to ensure deterministic end-to-end workflows.

How does DOM interrogation work across different navigation states?▼

DOM interrogation across different navigation states works by taking interactive snapshots of the page to extract specific element values, text, and attributes. You can scope searches and wait for UI states to target and extract machine-readable JSON data accurately at any navigation step.

Are there limitations when scoping searches and recording parallel browser sessions?▼

While parallel browser sessions and session recording are supported for later review, limitations arise if you do not use refs from interactive snapshots for actions. Relying on deterministic actions and proper wait conditions is necessary to avoid scoping errors during DOM interrogation.