agent-browser

Automates browser interactions to test web UIs, verify visual changes, and capture screenshots.

Updated Aug 14, 2026
One-click install
npx skills add https://github.com/tatran0195/emdash-with-docs --skill agent-browser-tatran0195
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/tatran0195/emdash-with-docs/tree/main/skills/agent-browser
Command: npx skills add https://github.com/tatran0195/emdash-with-docs --skill agent-browser-tatran0195

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Verifying frontend changes and testing user flows manually in a browser is slow and repetitive. This Skill gives AI agents a CLI to open pages, inspect the accessibility tree, interact with elements, and capture screenshots for automated verification. ## Core Features & Use Cases - Element Interaction via Refs: Snapshot pages to get interactive element refs, then click, fill, type, select, and check elements by reference. - Screenshots & Inspection: Capture viewport or full-page screenshots, read text/HTML/values, and check element visibility or state. - Sessions & Debugging: Persist login state with named sessions, view console logs and page errors, and run headed mode for debugging. - Use Case: After modifying a form in your Astro admin panel, open the page, snapshot interactive elements, fill the title and content fields, click save, and screenshot the result to confirm the change works. ## Quick Start Use the agent-browser skill to open http://localhost:4321, take an interactive snapshot, and verify the new post form renders correctly.

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 testing of web UI changes?▼

Use agent-browser to open the page URL, run snapshot with the -i flag to list interactive elements, then interact using refs like click @e5 or fill @e3. Finish with a screenshot to visually verify the result.

How to fill and submit a form with browser automation?▼

Take an interactive snapshot first to get element refs, then use fill to clear and type into fields, select for dropdowns, and check for checkboxes. Click the submit button ref and snapshot again to confirm submission.

Can browser automation sessions persist login state?▼

Yes, named sessions keep cookies and storage between commands. Pass --session mytest to each command or set the AGENT_BROWSER_SESSION environment variable to reuse authenticated state.

Why is my browser automation click not working?▼

The element ref may be stale or the page still loading. Re-run snapshot after navigation, add a wait command after actions that trigger loading, and use --headed mode plus console and errors commands to debug.

What are the limitations of accessibility-tree browser automation?▼

Refs change whenever the page updates, so snapshots must be refreshed after navigation or dynamic rendering. Screenshots larger than 2MB cannot be loaded by tools and must be resized with sips or ImageMagick first.