puppeteer

name	puppeteer
description	Use puppeteer to browse websites

Connection

curl -s --connect-timeout 2 http://127.0.0.1:9222/json/version | head -c 200

Empty / FAILED output means Chrome was not launched with --remote-debugging-port=9222 — stop and ask the user.

Resolve the WS endpoint (do not hardcode the UUID)

The browser UUID changes every launch. Always discover it:

WS=$(curl -s http://127.0.0.1:9222/json/version | node -e "let d='';process.stdin.on('data',c=>d+=c).on('end',()=>console.log(JSON.parse(d).webSocketDebuggerUrl))")
echo "$WS"

Connect and navigate

Run from /workspace so puppeteer-core resolves from the workspace node_modules. The heredoc form below uses tsx for top-level await; node --input-type=module works equivalently.

WS_ENDPOINT="$WS" tsx << 'EOF'
import puppeteer from "puppeteer-core";
const browser = await puppeteer.connect({
  browserWSEndpoint: process.env.WS_ENDPOINT,
  defaultViewport: null,
});
const page = (await browser.pages())[0] || await browser.newPage();
await page.goto("http://example.com:8080/", { waitUntil: "networkidle2", timeout: 20000 });
console.log(await page.title()); 
await browser.disconnect();        // never close() — keeps Chrome alive for next call
EOF