Search the web without any API key from a self-operated pool: Bing, 360, Baidu, Google, Naver, Yandex, and DuckDuckGo. Each search keeps at most two usable engines — the pool is routed by query language when no explicit pool is given, failed or irrelevant engines are back-filled within a bounded attempt budget, and results are relevance-ranked, quality-adjusted, and deduplicated by canonical URL. Each result includes a title, the real destination URL, a snippet, and its engine. Engines unreachable from the local network, or walled by captcha/JS/bot checks, are skipped and reported in an engine note; the remaining engines still answer.

Fetch one HTTP(S) URL anonymously and return the page as readable text: HTML is stripped to visible text with block structure preserved; text/markdown/JSON/XML pass through decoded. No credentials: browser User-Agent, up to five validated redirects, bounded output. Local, private, link-local, and reserved network targets are blocked, including after DNS resolution and redirects. The result is a status header (status code, final URL, content-type, truncation) followed by the content. Use it to read full page content for a URL, e.g. one found by web_search.