Skip to content
Get packs

Browser Automation with agent-browser

Lets the agent open web pages, read them, click, fill forms, and take screenshots via a browser CLI.

Source Download skill
Repository
nanocoai/nanoclaw
Stars
31,072
Licence
MIT
Files
1 file

What it does

This skill teaches an AI agent to control a real web browser from the command line. Once installed, the agent can open a website, read its interactive elements, click buttons, fill in forms, extract text or data, save screenshots or PDFs, and wait for pages to load. It is a practical way for an agent to research, check a web app, or gather information that is only available on a web page.

Great for

  • Developers who want an agent to test or check a web page or web app they are building
  • Researchers or analysts who need an agent to pull data from websites
  • People who want an agent to complete repetitive web form tasks

Skip it if

  • Sensitive logins or banking tasks, because the skill stores session state in files (auth.json) and fills in passwords, which creates credential exposure risk
  • Situations where the agent should not reach the internet, since the skill browses live websites
  • Users who expect a fully sandboxed, read-only research tool; the browser can click, submit forms and upload files

Try asking

  • “Open example.com/products, take a snapshot, and extract all product titles and links”
  • “Fill in the contact form on this URL and confirm the success message appears”
  • “Take a full-page screenshot of my staging site's homepage and save it as home.png”
  • “Log into my dashboard, save the session state, and check the latest report”
Step-by-step guide: how to use Browser Automation with agent-browserInstall, 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

  • Always ask the agent to re-take a snapshot after navigation so it works with current element references
  • Keep credentials out of chat prompts and avoid saving auth.json files in shared or synced folders
  • Use the bounded wait pattern from the skill for custom conditions so a stuck page cannot hang the agent

What's inside

  • SKILL.md

Redistributed unmodified from nanocoai/nanoclaw 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.