Skip to main content
A URL knowledge base connects an agent to live, real-time information from a website.
Unlike documents, URLs are not indexed — the agent fetches content from the site each time it is queried.

​
How does it work?

  1. Add a URL to your knowledge base list.
  2. Click the Test URL button to confirm the site is accessible.
    • Some sites may be inaccessible due to:
      • CAPTCHAs
      • Anti-bot restrictions
      • Other access barriers
  3. You can also test accessibility by asking the agent about the site and verifying its response.
  4. When the agent receives a question, it fetches the latest content in real time from the specified URL.

​
Best Practices

Real-Time Only
URLs pull content directly from the web; nothing is cached or pre-indexed. No File Links
Avoid using URLs that point to downloadable files like .pdf, .docx, etc.
Instead: Download the file and upload it via a Document Knowledge Base. Verification
Always test a new URL by asking the agent about it and validating the returned response for accuracy. Clarity
Use URLs that contain clean, static, text-based content.
Avoid heavily scripted or dynamic pages that make extraction difficult. Limiters
If you want your agent to only use certain parts of your site:
  • Use the exact URL to a specific page
  • Avoid linking just the home page unless necessary
⌘I