Fetch a web page (Japanese-web-native) as low-impact, token-efficient Markdown. Built-in robots.txt compliance, rate limiting, and caching. mode: auto (default; quality score picks Markdown or screenshot) / markdown / outline (heading summary) / screenshot. Page 1 can come back as cache: unchanged or cache: diff (changed sections only, each without its child sections) instead of the body, but only when all of these hold: this server process has already returned that URL's Markdown body, amenbo's cached copy has expired (15 min by default; a refetch within 15 min, or a site that answers 304, returns the full text), the site resent the whole page, and the call has no selector and no force_full. The comparison copy may have been stored by another session or the CLI. Both responses carry total_tokens (the size of a full refetch) and a hint: if you have not read the whole content yourself, refetch with force_full: true. The CLI never returns them. The response also lists markdown_alternate: and llms_txt: when the page advertises them (amenbo does not fetch them itself). PDF URLs are handled automatically; other non-HTML responses (CSV, JSON, text/markdown, etc.) return a handoff with metadata, a text preview (paginated by page) and a curl command for the whole file. mode: screenshot is the retry for pages whose Markdown came back poor; when the layout itself is the information or you need viewport/scale control, use the screenshot tool instead.

Enumerate URLs under a site or page without fetching each one. Tries, in this order and regardless of what is cached: sitemap.xml (from robots.txt or the conventional path; gzip is unpacked up to the body-size limit and a warning: line is added when it cannot be read), then the llms.txt the page advertises, then RSS/Atom feeds, then falls back to links found in the page itself. The response header reports which source was used (source:), the count:, and (truncated) when more than 200 links were cut. Each link is one line, - <title> — <url> (title omitted when unknown; for llms.txt it is name: description). filter narrows by substring or * glob against URL and link text; an empty result states whether the filter removed everything or the source had no links. Each source is cached like a page (15 minutes by default, up to 24 hours when Cache-Control max-age says so), so repeating the call with another filter does not hit the site again. Use it to find the right page before fetch. It does not return page content, reads the first 3 child sitemaps of a sitemap index in order and stops reading further children once 200 links are collected (filter is applied afterwards, so a URL found only in an unread child cannot be matched), and never guesses an llms.txt location.

Render a page in a headless browser and return it as up to 10 tiled PNG images (1280px wide by default; fullPage: false returns only the first viewport). Pages taller than 10 tiles are cut off and the response says truncated: true. Use it when the layout or an image is the information (pricing tables, banners, visual QA) or when you need explicit viewport/scale control; scale: 0.5 roughly halves image tokens. For text content use fetch. Built-in robots.txt compliance, rate limiting, and caching.