Skip to content
Get packs

Browser Harness

Lets the agent drive a real Chrome browser: click, type, navigate, and use your logged-in sessions.

Source Download skill
Repository
browser-use/browser-harness
Stars
18,815
Licence
MIT
Files
6 files

What it does

Once installed, the agent can control a real Chrome or Chromium browser through the Chrome DevTools Protocol, performing clicks, typing, navigation and screenshots on pages that need interaction, JavaScript rendering, your logged-in session, or that block simple fetches. For plain reads of public pages or APIs, the skill tells the agent to use curl or a basic fetch instead, and to open a browser only when those fail. It also manages browser tabs carefully, reusing a single local connection and avoiding stealing focus from your windows.

Great for

  • Developers automating web tasks that need clicks, forms or logged-in access
  • People whose agent must operate on JavaScript-heavy or bot-protected sites
  • Users who want background browser control without Chrome jumping to the foreground

Skip it if

  • Simple reads of public pages or APIs, where curl or a plain fetch is cheaper and safer (the skill itself says so)
  • Sensitive accounts such as banking or email, since the agent acts with your logged-in session and can change data, which is risky without close supervision
  • Environments where Chrome remote debugging cannot be enabled or the user cannot approve connection prompts

Try asking

  • “Log into my dashboard and download this month's invoice CSV”
  • “Fill out the registration form on this site and take a screenshot of the confirmation page”
  • “Click through the search results on this page and collect the prices”
  • “Check the status of my order on this retailer's site using my logged-in account”
Step-by-step guide: how to use Browser HarnessInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 6 original files, the MIT licence and an attribution note.

    browser-harness.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

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

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip browser-harness.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 try curl or a plain fetch first for public content so the browser is only used when needed
  • Keep sensitive sessions supervised and review what the agent clicked or submitted
  • If connection problems occur, run browser-harness --doctor and follow install.md

What's inside

  • AGENTS.md
  • CONTRIBUTING.md
  • install.md
  • pyproject.toml
  • README.md
  • SKILL.md

Redistributed unmodified from browser-use/browser-harness 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.