agent-browser

Automate browser navigation, form filling, screenshots, and data extraction via a CLI.

1|Updated May 10, 2026
One-click install
npx skills add https://github.com/Tgoldi/claude-skills --skill agent-browser-tgoldi
Or copy as Structured Prompt for Agentâ–¼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/Tgoldi/claude-skills/tree/main/agent-browser
Command: npx skills add https://github.com/Tgoldi/claude-skills --skill agent-browser-tgoldi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser, and includes references (resource) components.

What problem does it solve? Manually interacting with websites for testing, scraping, or repetitive form entry is slow and error-prone. This Skill gives an AI agent full programmatic control of a Chrome/Chromium browser through the agent-browser CLI, turning multi-step web tasks into simple shell commands. ## Core Features & Use Cases - Snapshot-and-Interact Workflow: Get compact accessibility-tree snapshots with element refs (@e1, @e2), then click, fill, select, and scroll using those refs instead of fragile CSS selectors. - Authentication & Session Management: Save and restore login state, use persistent profiles, named sessions, or an encrypted auth vault so automated runs stay logged in. - Capture & Verification: Take annotated screenshots, export PDFs, record video, diff page states, and inspect network requests for testing and debugging. - Use Case: Ask the agent to log into a web app, navigate to a dashboard, extract a table of data, and save a full-page screenshot as evidence — all in one automated run. ## Quick Start Ask the agent to open a website with agent-browser, snapshot the page, and fill out and submit the login form.

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 actions with agent-browser?â–¼

Open a URL with agent-browser open, run snapshot -i to get element refs like @e1, then use click, fill, and select with those refs. Re-snapshot after any navigation or DOM change because refs are invalidated when the page changes.

How do I keep a browser session logged in between runs?â–¼

Use the auth vault with agent-browser auth save and auth login, or save state with state save and reload it with state load. Named sessions via --session-name also auto-save and restore cookies and localStorage.

Can agent-browser connect to an existing Chrome instance?â–¼

Yes. Use --auto-connect to discover a running Chrome with remote debugging enabled, or --cdp 9222 to connect to an explicit CDP port. This lets you reuse an already-authenticated browser session.

Does agent-browser support mobile or iOS testing?â–¼

Yes. With -p ios and --device you can drive Safari on iOS simulators, using tap, swipe, and fill commands. It requires macOS with Xcode and Appium with the xcuitest driver installed.

Why do my element refs stop working after clicking a button?â–¼

Refs are invalidated whenever the page changes, including navigation, form submissions, and dynamic content updates. Run snapshot -i again after any such action to get fresh refs before interacting further.

How do I run JavaScript safely in the browser with agent-browser?â–¼

Use eval --stdin with a heredoc or eval -b with base64-encoded scripts to avoid shell quoting corruption. Simple single-line expressions can use regular eval with single quotes.