nc://browserless

Browserless: renders pages, screenshots them and prints them to PDF in a stealth cloud browser. A Nodecode rewrite of browserless/browserless-mcp by browserless.

Browserless is a hosted Chromium: one HTTP request renders one page on their machines, with stealth on, optionally through a residential IP, with ad blocking and captcha solving. This cell gives the model three tools that use it, so a script-heavy, login-gated or datacenter-blocked page loads without a browser on your machine. Everything runs on Browserless' side; the cell installs nothing and starts nothing.

Get it

Tell the agent "add the browserless cell", or run nodecode add nc://browserless.

Set it up

Get an API token from https://www.browserless.io/account, then name it in the cell's section — in /settings, on the settings page, or by asking the agent to set browserless.token_env:

{
  "browserless": {
    // the environment variable holding your token ...
    "token_env": "BROWSERLESS_TOKEN",
    // ... or a file holding it (set one, not both)
    "token_file": "",
    // the region endpoint your account answers on
    "base_url": "https://production-sfo.browserless.io"
  }
}

stealth is on by default. proxy = "residential" (with proxy_country, e.g. "us") routes the page through a residential IP, which uses more of your plan's units; solve_captchas and block_ads are off by default. timeout_seconds bounds one page and one request, and content_chars caps how much page text a tool answers.

Use it

The model calls the tools itself; you can also call them from a snippet:

• browserless_fetch — a page as its words (default) or its rendered HTML, with an optional wait_for_selector and wait_ms for content that fills in late.
• browserless_screenshot — a PNG of the page (optionally full-page, with a viewport size), answered as a path to look at with (look PATH).
• browserless_pdf — the page printed to A4, answered as a path.

/browserless shows what the cell will do — endpoint, whether the token is set, stealth, proxy, timeout — and /browserless <url> renders one page and shows its words, which is the quickest way to see whether your token and proxy settings work.

Pages are rendered on Browserless, so a localhost or LAN address is not reachable from there: use a local fetch for this machine. A page longer than content_chars is cut, and the answer says how much there was.