web-search

To fetch real-time web information, use webSearch to find relevant URLs and webFetch to extract page content as markdown. This retrieves current data for research, API documentation lookups, and fact verification beyond training data limits.

You can run multi-query concurrent web searches by passing up to 10 related queries at once. This allows you to compare documentation, policy updates, or version differences across vendors simultaneously for efficient real-time research.

The best way to verify facts and find current API rate limits is using web search to locate relevant pages and fetching their content as markdown. This ensures you get up-to-date, reliable answers when training data is insufficient.

Web search and URL fetching have known limitations with blocked domains and inaccessible paywalled content. The tool cannot retrieve or extract text from these restricted sources, so you should seek accessible URLs for markdown extraction.

To get up-to-date info from the web when training data is outdated, execute a webSearch query for ranked results, then use webFetch on selected URLs. This retrieves current page content as markdown for reliable, real-time answers.