Web retrieval & search

Web capabilities separate direct source retrieval from discovery and from stateful browser interaction.

Fetch

fetch_url retrieves a response. fetch_readable_url extracts bounded reader-oriented content. Use them when page interaction and client-side state are unnecessary.

Search

web_search discovers current external sources. Prefer primary documentation, official specifications, and corroborated claims; retain links with the answer.

Escalate to browser

Use browser automation for JavaScript applications, authentication, forms, navigation, screenshots, or state that a static fetch cannot observe.