Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

agent-browser Automation Guide

Teaches Claude to drive a real Chrome browser through the agent-browser CLI using accessibility-tree snapshots and compact @eN refs for clicking, filling, extracting and screenshotting.

AutomationIntermediate★ 1,161⑂ 109AI score 10/10Last updated: Sep 29, 2026

What it does

  • Establishes the core loop for browser work: open → snapshot -i → act on @eN refs → re-snapshot after any page change.
  • Uses accessibility-tree snapshots (~200-400 tokens) instead of raw HTML parsing, plus semantic find role/text/label locators and CSS as a fallback.
  • Covers reading pages (read, get text/attr/value), interaction (click, hover, fill, type, select, check, upload, drag, scroll), screenshots (full-page, annotated), tabs, iframes and dialogs.
  • Documents login flows with the auth vault and credential-provider plugins, session persistence via --restore, and parallel isolated browsers via --session.
  • Adds network mocking/HAR capture, workflow video recording, axe-core accessibility audits, and React introspection (component tree, re-render profiling, Web Vitals).
  • Includes a strong waiting strategy plus a troubleshooting section for stale refs, overlays swallowing clicks, custom inputs ignoring key events, cross-origin iframes and WebGPU black screenshots.

Who it's for

  • Developers and QA engineers automating repeatable web-app checks or bug hunts.
  • Analysts scraping data from sites that require a logged-in session.
  • Frontend engineers who want automatic before/after screenshots.
  • React teams profiling re-renders and Core Web Vitals.

Examples

  1. "Log into this app and screenshot the dashboard" → snapshot to find email/password refs, fill and submit, wait --url "**/dashboard", then capture.
  2. "Turn this pricing table into JSON" → pipe a heredoc into eval --stdin to map table rows into objects.
  3. "Test the chat feature with two users at once" → run --session a and --session b as fully isolated browsers with separate cookies.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/fcakyon/claude-codex-settings/HEAD/plugins/agent-browser/skills/agent-browser/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the plugins/agent-browser/skills/agent-browser folder from the GitHub repo fcakyon/claude-codex-settings into my ~/.claude/skills/fcakyon-agent-browser/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/fcakyon/claude-codex-settings.git /tmp/claude-codex-settings && mkdir -p ~/.claude/skills && cp -r /tmp/claude-codex-settings/plugins/agent-browser/skills/agent-browser ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Confirm Node.js is installed: node -v.
  2. Install the CLI and its browser: npm i -g agent-browser && agent-browser install (on Linux use agent-browser install --with-deps).
  3. Clone the skill repository: git clone https://github.com/fcakyon/claude-codex-settings.git
  4. Copy the skill folder into place: mkdir -p ~/.claude/skills && cp -r claude-codex-settings/plugins/agent-browser/skills/agent-browser ~/.claude/skills/
  5. Run agent-browser doctor to verify Chrome, daemons and config (install ffmpeg too if you plan to record video).
  6. Restart Claude Code and try a prompt like "open example.com and take a screenshot" to confirm the skill triggers.
  7. Run agent-browser close when finished to shut the browser session down.
View source on GitHub ↗License: Apache-2.0