Luni Studio
A prompt, with or without a URL. Review the preview, save the rules, and collect the same data again on your schedule.
Generate from a page
Give Studio a public URL and describe the records you need. Without a URL, describe what you want, such as “a scraper for Hacker News”, and Studio chooses a public page and tells you which. It samples the page, proposes bounded extraction rules, and tests them against that fetch so you can review records, missing fields, and the rules before saving.
Choosing a page from your request adds no charge. A validated generation preview costs one credit. Saving adds no charge. Later runs reuse the saved rules at one credit per useful successful page. Page, Article, and Product templates stay available when AI generation is off.
Run once or on a schedule
Run scrapers manually any time, or authorize every 24 hours or every 7 days. Schedules start off; turn one on to see the next run time, or turn it off to stop future work. Deleting a scraper stops its schedule.
Scheduled runs reserve credits first. If the workspace is out of credits or access, new runs do not start. Admitted runs keep their original config and stay in request history.
Inspect results
Studio shows run state, settled credits, saved output, and failure diagnostics. Results are kept for seven days. Editing a scraper changes future runs without rewriting past operations.
Previews cover one fetched page. Pagination, automatic repair, and interactive logins are outside this flow. Retrieval uses public HTTP, so pages that need a browser to render may not preview well.