Forty client reports, forty PDFs, one browser URL

Month-end. Forty retainer clients each get a performance report. The report is a page your exec script renders from their data: /reports/acme?month=2026-07. It's already live; some clients just read it there.

The ones who want a PDF get it from the same page. A loop hits GET /api/v1/browser/pdf?browser_id=0&url=…/reports/acme&format=A4&printBackground=true, and hoody-browser navigates, waits for the charts, and returns the file. Forty calls, forty PDFs, written into a dated folder on hoody-files where the account managers pick them up.

There's no headless-Chrome Lambda to keep warm, no Puppeteer version to chase, no report-generation service. The browser is one of the Kit programs on the same flat-rate box as the data. Every client is a URL; every PDF is that URL with a different host in front.

The PDF stops being a pipeline and becomes a query string.