---
title: "Import from a live docs site"
url: "/help/buuk/import-from-live-site"
updated: "2026-10-22T11:13:40.505Z"
---
# Import from a live docs site

If your documentation already lives on a public site, point Buuk at it. Buuk crawls the site, converts each page to editor blocks, and lands them as drafts.

ReadMe and Mintlify sites are detected by name and get platform-specific conversion. Everything else — GitBook, Docusaurus, Zendesk Guide, or a self-hosted docs folder — works the same generic way, because discovery does not depend on the platform: if pages are linked, sitemapped, or listed in llms.txt, they import.

## How discovery works

Buuk reads robots.txt first, then builds the page list with the first method that works:

1. `llms.txt` — section headers become sections and list items become pages. On ReadMe and Mintlify this usually points at a clean markdown version of each page.
2. `sitemap.xml` — follows sitemap indexes through up to three levels.
3. Links — a same-origin crawl from your start URL, following page links up to four levels deep and staying within the path the start URL is under.

> [!NOTE]
> robots.txt is respected. A URL it disallows is never fetched, and the crawl delay it declares sets the pace. If robots.txt cannot be read at all, importing is disabled for that site.

## Run an import

1. **Open the importer**
   Go to **Settings → Import content** and choose **From a website**. If the card is not listed, website import is switched off for your workspace — ask an administrator to enable it.

2. **Set the target**
   Paste the **Documentation URL**. Pick a **Platform** (**Detect automatically**, ReadMe, Mintlify, or Other docs site) and a **Page cap** — up to 500 pages per import, 200 by default.

3. **Confirm ownership**
   Tick **I own this content or have permission to import it**. The import refuses to run without it. Only copy documentation you are allowed to use.

4. **Check the site**
   Select **Check site** to preview what was found: the platform, how the pages were discovered, whether robots.txt allows the start URL, and the list of pages.

5. **Import as drafts**
   Select **Import as drafts**. The import runs in the background — you can close the tab and reopen this page to watch progress. Cancelling keeps every page fetched so far.

6. **Review and publish**
   When it finishes, download the report (CSV) to see every page with what happened to it. Fix slugs and section placement in Articles, then publish each page when it reads right.

## How the crawl behaves

- About one request per second to the site's host by default; a robots.txt Crawl-delay raises that, with a five-second ceiling.
- Starting imports repeatedly in a row is rate limited. Wait a moment and try again if you hit the limit.

Mintlify-style components (steps, callouts, code groups) are converted to plain markdown equivalents during import. Anything that cannot be converted is dropped and reported, not silently mangled.

## Re-importing the same site

Each page's canonical URL is its identity. Run the same import again and matched drafts update in place — nothing is duplicated, and sections keep the order they were first given.

A page you already published is skipped rather than overwritten, so a re-run never changes a live page. Nothing publishes on its own, ever.

## Related

- [Migration overview](./migration-overview)
- [Import markdown](./import-markdown)
- [Slugs and redirects](./slugs-and-redirects)
- [Organize sections and folders](./organize-sections-and-folders)
- [Publish your help center](./publish-your-help-center)