pi-web-browse

Web search and content extraction skill for pi. Search the web and fetch pages via a real headless browser (CDP).

Works on Linux, macOS, and Windows.

Features

• 🔍 Web Search - Search via Google (falls back to DuckDuckGo if blocked)
• 🌐 Page Fetching - Extract readable content from any URL
• 🤖 Bot Protection Bypass - Handles JS challenges, Cloudflare, etc.
• 🚀 Persistent Daemon - Warm browser session for fast subsequent requests
• 🖥️ Cross-Platform - Auto-detects Chrome, Brave, Edge, Chromium

Install

pi install npm:@ogulcancelik/pi-web-browse

Or via git:

pi install github.com/ogulcancelik/pi-web-browse

(Optional, try without installing):

pi -e npm:@ogulcancelik/pi-web-browse

After first use, the agent will guide you through setup.

Usage

The agent will automatically use this skill when you ask it to search the web or fetch page content.

You can also invoke it directly:

/skill:web-browse "rust async runtime"

Configuration

Environment variables (all optional):

Variable	Description	Default
WEB_BROWSE_BROWSER_BIN	Browser binary path	Auto-detected, prefers Chrome
WEB_BROWSE_USER_AGENT	Override User-Agent string	Auto-derived from detected browser + OS
WEB_BROWSE_DAEMON_PORT	Daemon HTTP port	9377
WEB_BROWSE_CDP_PORT	Chrome DevTools port	9225
WEB_BROWSE_DEBUG_DUMP	Save debug files on failure	off

By default, the hidden profile is browser-specific (for example ~/.config/web-browse-cdp-profile-chrome or ~/.config/web-browse-cdp-profile-brave) so updates do not try to reuse the same hidden profile across different Chromium-family browsers.

Browser Detection

The skill auto-detects browsers in common locations and now prefers Chrome first, because Google Search is less likely to challenge headless Chrome than headless Brave on some setups.

• Linux: google-chrome, google-chrome-stable, brave, brave-browser, chromium (from PATH)
• macOS: Google Chrome, Google Chrome Canary, Brave Browser, Edge, Chromium (in /Applications)
• Windows: Chrome, Brave, Edge, Chromium (Program Files, LocalAppData)

If you want Brave anyway, set WEB_BROWSE_BROWSER_BIN or pass --browser-bin.

How It Works

1. Search - Uses Google via a persistent headless browser session. If Google blocks the request, it fails fast and falls back to DuckDuckGo.
2. Fetch - Opens URLs in the same hidden browser session, waits for JS, and extracts readable content.
3. Daemon - Keeps a warm, browser-specific hidden profile/session alive for speed and bot-protection resilience without touching your normal browser profile.

License

MIT