browser-use

Use this skill when a task genuinely needs browser automation through PilotDeck's built-in browser-use plugin, especially for local Web UI smoke tests, screenshots, navigation, clicking, typing, and DOM inspection. For static pages, batch scraping, API responses, or plain text extraction, prefer ordinary HTTP/file tools first.

Availability

PilotDeck ships the browser-use plugin, which runs @playwright/mcp with Chromium. Before installing anything, first try to use the existing browser/MCP setup or check the browser cache. If the browser is already present or browser-use launches successfully, do not reinstall it.

Check whether the browser is already installed:

if ls "$HOME/Library/Caches/ms-playwright"/mcp-chrome-for-testing-* >/dev/null 2>&1 || \
   ls "$HOME/.cache/ms-playwright"/mcp-chrome-for-testing-* >/dev/null 2>&1; then
  echo "Chrome for Testing is already installed; no reinstall needed."
else
  echo "Chrome for Testing is not installed yet."
fi

If that check reports an existing Chrome for Testing cache, proceed with browser automation directly. Repeated browser downloads are slow, brittle on restricted networks, and unnecessary when the cache is already populated.

Only install the browser when all of the following are true:

• The task truly requires an interactive browser rather than HTTP, curl, requests, or file parsing.
• The cache check shows Chrome for Testing is missing, or an actual browser-use launch failed because the browser executable is missing.
• The environment has network/proxy access suitable for downloading browser binaries.

The one-line installer uses the same check. If Chrome for Testing is already present, it prints Chrome for Testing already installed and does not download it again.

Install it from a PilotDeck source checkout or installed app directory: