Skip to Content
SourcesWebsite crawl

Website crawl

Website crawl indexes published HTML pages. Use it when your docs are already live and the rendered site is the easiest source of truth.

Create a crawl

  1. Open Project -> Sources.
  2. Choose Website crawl.
  3. Enter the root URL and set path scope (all pages, include only, or exclude paths).
  4. Run Preview to see discovered URLs and deselect pages you do not want.
  5. Start the job.

The job runs in the background and shows progress on the Sources page.

Discovery

Knoku discovers pages in this order:

  1. sitemap.xml (and nested sitemaps)
  2. sitemap entries listed in robots.txt
  3. links from the root page
  4. vendor /map endpoints and deeper crawl when sitemap and root links are not enough

robots.txt is always respected. Static assets such as images, fonts, and build artifacts are skipped.

Options

OptionWhat it controls
Root URLStarting URL for discovery
Path scopeAll pages, include-only paths, or exclude paths
Page selectionDeselect URLs after preview before indexing

There is no dashboard control for max pages or robots.txt — robots are always honored, and page caps are applied automatically during discovery.

Conflict handling is not a crawl-start option in the dashboard. New crawls always flag clashes instead of auto-overwriting.

After indexing

Crawled pages become web_page documents. Citations use the live page URL.

Use Re-crawl page from a document row when a single page changes. Use Re-crawl on the crawl source when the site structure changes.

Conflicts

If a crawled URL matches a document from GitHub repo sync or another source, the crawl job marks those pages as conflicts instead of silently overwriting.

When a job finishes with conflicts, open Project -> Sources, find the crawl job, and click resolve. Choose one action for every conflict in that job:

  • Skip — keep the existing document; drop the crawled page.
  • Replace — soft-delete the existing document and ingest the crawled page.

The CLI and Public API still accept conflict_resolution: error | replace | skip on each request if you automate crawls outside the dashboard.

Plan

Website crawl is available on all plans. Business+ projects can enable automatic crawl refresh.

Last updated on