Any page could hold URLS to pages that do have holders on there, also pages that do have holders. It might be a good idea to do this in the same trajectory as where we extract holders, as we can take another turn with different structured output, making use of the cached input tokens and existing context. I'm also curious whether we can use sitemap.xml as a signal (or spider?) for what pages are interesting to look at.
Any page could hold URLS to pages that do have holders on there, also pages that do have holders. It might be a good idea to do this in the same trajectory as where we extract holders, as we can take another turn with different structured output, making use of the cached input tokens and existing context. I'm also curious whether we can use sitemap.xml as a signal (or spider?) for what pages are interesting to look at.