WebSearchTool

Searches the web by scraping DuckDuckGo's HTML endpoint (html.duckduckgo.com) with OkHttp and parsing results with Jsoup. Reaches the network (openWorldHint); requires INTERNET (always-granted). Results depend on DuckDuckGo's HTML structure (.result / .result__a / .result__snippet) — a markup change can yield zero parsed results without an error. Any HTTP or parse failure is caught and returned as a ToolResult.error; a non-2xx status returns an error too.

Output map: query (echoed), results (List<Map> each with title/url/snippet), result_count (Int).