Worked examples
Four workflows. Each one was built, run on a robot, and checked against the file it produced — the row counts and console output on these pages are from those runs, not from a description of what should happen.
Several of them failed first. Those failures are on the pages too, because the wrong version is usually the more useful half: it is the version you are about to write.
| Example | What it shows |
|---|---|
| API to CSV | HTTP, a keyed in-memory table, CSV export |
| Scrape a catalogue page | Element arrays, joining them by index, Excel export |
| Retry and recover | RetryOnError inside try, so one bad record does not end the batch |
| Page to PDF | Printing a page, and why it needs a headless browser |
Each page gives the whole workflow as DSL. Paste it into the designer's Code tab, or hand it to an assistant over MCP.
They run against public endpoints — jsonplaceholder.typicode.com and
books.toscrape.com, both of which exist to be called — so you can run them unchanged
before pointing them at anything of yours.