Crawl a single page
Add one public URL as knowledge when a full website crawl would be too broad.
Open Sources → Content and select the Pages tab to index one public URL. This is useful for a specific returns policy, release note, or help article when you want to avoid discovering the surrounding site. If the information is a short, stable answer, a Q&A pair may be even more precise.
Add the page
Under Add Page, enter the Page URL and select Fetch Page. AgentStack adds the HTTPS scheme when the URL has none. A successful click queues the source; it does not mean the page is ready for retrieval.
The Crawled Pages table shows the URL, status, and controls. Wait for Completed before testing. Pending, Crawling, and Processing mean work is still in progress; Failed means the page did not become usable, so check the URL’s public accessibility and retry or replace it. Delete a page that no longer belongs in the agent’s knowledge.
Test what was indexed
After the page completes, ask a question whose answer appears in its text in Playground. Then ask a paraphrase. If the result is missing, confirm that the page’s actual body contains the fact, not merely a navigation label or dynamically loaded element.
Understand later retraining
A manually added page is stored under a website grouping. A later retrain can rediscover from that grouping URL using the current plan page limit; it is not guaranteed to refresh only that one exact URL. Use the website crawl guide when you intend ongoing discovery across a documentation area, and review Retrain knowledge before enabling regular refreshes.