Answers

How to Import Your Existing Help Site with Site Import

A walkthrough of DocsKoala's Site Import feature: point it at a URL, watch it crawl and draft articles from your existing docs, and review everything before anything publishes.

Written by The DocsKoala Team

If you already have help content somewhere, a docs site, an FAQ page, a support section on your marketing site, Site Import gives you a starting point instead of writing from a blank help center. Point it at a URL, and DocsKoala crawls the site, works out which pages actually look like help content, and drafts articles from what it finds.

It's a one-time bootstrap, not a live sync: it doesn't watch your old site for changes afterward. And like every other drafting source in DocsKoala, nothing it produces goes live on its own, every draft lands in the Review Inbox for a human to check first.

Direct answer

To use Site Import, go to Integrations in your DocsKoala dashboard (Startup plan and up, admin role required), find the Site Import panel, and enter the URL of your existing docs or help site, then click Import. DocsKoala crawls same-origin pages starting from that URL (up to 50 pages, 3 levels deep, respecting robots.txt), prioritizes paths that look like documentation (/docs, /help, /support, /faq, and similar) over general marketing pages, and drafts an article for each page it keeps. Progress updates live through Queued, Crawling, Drafting, and Done status, and every resulting draft appears in your Review Inbox for approval, exactly like a draft from any other source. Only one import can run per org at a time.

Before you start

  • Plan: Site Import requires the Startup plan or above.
  • Role: you need an admin (or owner) role in the org to start an import.
  • One at a time: only one active import can run per org. Wait for the current one to finish (or cancel it) before starting another.
  • What to point it at: the URL of your existing docs, help center, or FAQ site, the page itself becomes the crawl's starting point, and the crawler stays on that same domain.

Running an import

  • Go to Integrations in your dashboard and find the Site Import panel.
  • Enter the URL of your existing site (e.g. https://docs.yoursite.com) and click Import.
  • The panel shows a live-updating status: Queued → Crawling → Drafting → Done (or Failed), refreshing automatically every few seconds so you don't need to reload the page.
  • You can cancel a run in progress if you started it by mistake or pointed it at the wrong URL.
  • Once it reaches Done, open the Review Inbox to see everything it drafted.

What the crawler actually does

This is deliberately conservative, not a full-site scrape. Knowing the actual limits helps you predict what will and won't get picked up.

  • Same-origin only. It never follows a link off the domain you gave it.
  • Respects `robots.txt`. Anything your site's robots rules disallow is skipped.
  • Up to 50 pages, 3 levels deep. This is a bootstrap, not a full mirror, large sites won't be imported in full on one run.
  • Also checks `/sitemap.xml` at your site's root as an additional source of same-origin URLs to consider, on top of following links from the starting page.
  • Prioritizes docs-shaped paths. Pages under paths like /docs, /help, /support, /faq, /guide, /tutorial, /changelog, /release-notes, or /api-docs are crawled first. Only if those yield fewer than 10 pages does it widen to secondary paths like /pricing, /features, /product, or /getting-started. General marketing, blog, and about/legal pages are skipped entirely, they're not what a help center bootstrap needs.

From crawled page to draft article

  • Each kept page is classified by what it looks like: docs, FAQ, support, blog, or general marketing content, based on its URL path.
  • Navigation, headers, footers, and other non-content chrome are stripped out before anything is drafted, only the actual page content is considered.
  • Long pages are split into sections before drafting, so a single sprawling page doesn't become one unreadable article.
  • Each section goes through the same draft-then-critique pass every other DocsKoala draft goes through, not a raw copy-paste of the original page.
  • Every resulting draft is tagged with its source (site import) and lands in the Review Inbox, same as a draft from a merged PR or a closed support ticket.

Reading the results

  • Page count shown in the run history tells you how many pages were actually kept and drafted, not how many were merely visited.
  • A "low confidence" warning means the crawl didn't find pages that clearly look like documentation, help, or FAQ content. Worth a manual look at what it did find, you may need to point it at a more specific URL (a /docs or /help subpath rather than your homepage).
  • Nothing publishes automatically. Every draft needs the same one-click human approval as any other draft in DocsKoala, review the imported drafts before approving, exactly like you would a PR-generated draft.

Frequently asked questions

Does Site Import keep syncing with my old site after the first run?

No, it's a one-time crawl. It doesn't watch your previous site for future changes, it's meant to give you a starting set of drafts, not an ongoing sync between two sites.

Will Site Import import my entire site?

No. It's capped at 50 pages and 3 levels deep per run, and it prioritizes pages that look like documentation over general marketing content. For a large docs site, expect a strong starting subset, not a full mirror.

Does it respect my site's robots.txt?

Yes, any path your site's robots rules disallow is skipped by the crawler.

Can it import pages that require a login?

No, the crawler only fetches publicly reachable pages on the domain you give it. It doesn't accept login credentials or crawl behind an authentication wall.

What happens to the drafts it creates?

They land in your Review Inbox exactly like drafts from any other source, GitHub, a closed support ticket, and so on. Nothing publishes until you approve it.

Can I run more than one import at once?

No, only one active import is allowed per org at a time. Let the current run finish or cancel it before starting another.

Conclusion

  • Site Import is a one-time bootstrap: point it at your existing docs or help site, it crawls up to 50 same-origin pages (prioritizing docs-shaped paths), and drafts an article per page it keeps.
  • It respects robots.txt, stays same-origin, and every URL it fetches is checked against an SSRF guard first.
  • Every draft lands in the Review Inbox like any other DocsKoala draft, nothing from this feature publishes without a human approving it.
  • Requires the Startup plan or above and an admin role to run.

The self-updating help center

Never write another stale help article.

DocsKoala watches your merged PRs, drafts the doc updates, and waits for your one-click approval. Start your 7-day trial and let it write your first few articles.

Start free trialSee how it works

Keep exploring

More answers from DocsKoala

How to Connect Every DocsKoala Integration: The Full Setup GuideComplete setup guide for every DocsKoala integration: GitHub, Zendesk, Intercom, Help Scout, Slack, Linear, Shopify, Telegram, email, custom webhooks, Site Import, MCP, and the public REST API.How to Move Your Help Center to a Self-Updating PlatformA practical guide to migrating your help center to a self-updating platform: preserve URLs and SEO, keep redirects, and pick a tool that maintains docs automatically.What Is a Review Inbox for AI-Generated Docs?A review inbox is where AI-drafted doc updates wait for human approval before publishing. What it does, why it exists, and how approval works.