Skip to content
Get packs

Agent Browser Automation

Lets an agent open websites, fill forms, click, take screenshots and extract page data via a CLI

Source Download skill
Repository
shanraisshan/claude-code-best-practice
Stars
68,452
Licence
MIT
Files
1 file

What it does

Once installed, the agent can control a real web browser through a command-line tool called agent-browser. It opens pages, reads the clickable elements on them, fills in forms, clicks buttons, waits for pages to load, takes screenshots or PDFs, and pulls text out of pages. It also supports saving login state, running several browser sessions at once, and driving an iOS simulator for mobile Safari testing.

Great for

  • Developers who want an agent to test their web app through the browser
  • People who need to repeat routine web tasks like filling out the same form or checking a page
  • Data gatherers who want an agent to pull text from specific pages

Skip it if

  • Logging into sensitive accounts such as banking or email, since saved login state (auth.json) is stored as a plain file and the agent would act with your credentials (risk: credential exposure)
  • Scraping sites whose terms of service forbid automation, since the skill includes proxy and rotating-proxy guidance that could be used to evade restrictions (risk: legal and ethical)
  • Users without the agent-browser CLI installed, since the skill only describes commands and cannot run a browser by itself

Try asking

  • “Open example.com/signup and fill in the form with my test details, then take a screenshot of the result”
  • “Log into my staging dashboard, save the session, and check that the reports page loads”
  • “Scrape the product names and prices from this page into a text file”
  • “Test this web app's checkout flow in the browser and tell me what breaks”
Step-by-step guide: how to use Agent Browser AutomationInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    agent-browser.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip agent-browser.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip agent-browser.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Ask the agent to re-snapshot after any page change, since element references go stale after navigation or DOM updates
  • Pass credentials through environment variables rather than typing them into chat, and delete any saved auth.json when finished
  • Use --headed mode to watch the browser while debugging a flow

What's inside

  • SKILL.md

Redistributed unmodified from shanraisshan/claude-code-best-practice under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.