Read a whole published page in one call

With Agno's Knowledge.read_full_page or aread_full_page, an agent that found the right page through search no longer walks it with read_page, one slice at a time, following next_offset to the end. Both methods make a single SQL read against one read-only snapshot, so the text can't change partway through.

hit = knowledge.search_pages("How do agents use tools?").results[0]
 
body = knowledge.read_full_page(hit.path, revision=hit.revision, max_chars=24_000)
if body is None:
    ...  # longer than the budget, so keep the search excerpts

Pass the revision from a search hit to pin the read. If someone republished the page after that search, read_full_page raises PageChanged. A page longer than max_chars returns None, and you can fall back to the excerpts you already have. A missing page raises PageNotFound. A read that runs past its timeout raises PageError. The defaults are 24,000 characters and 2 seconds.

read_full_page shares the eight-slot worker pool that Agno's other page reads use.

See the cookbook, and learn more about published pages in the documentation.