RSS Feed Finder for Page HTML

An RSS feed finder should show the clue before claiming the feed

A page can advertise a syndication feed with a <link rel="alternate"> element whose type names RSS, Atom or JSON Feed. It can also expose a normal link labeled RSS, Atom or subscribe. Those are useful discovery facts, but they are still only facts about the HTML page. The target may have moved, require access, return an error, contain ordinary HTML or be malformed as a feed.

That is why every result here keeps evidence and network status apart. A typed alternate declaration is stronger source evidence than a feed-looking anchor, but even its MIME type is advisory under the HTML Living Standard. The report says what the author declared and never converts that declaration into a green “valid” badge without reading the target.

Find an RSS feed on the current page without crawling the website

The bookmarklet reads the document already loaded in your browser. It takes typed alternate declarations from the head and visible anchors whose resolved HTTP or HTTPS URL or short label contains a bounded RSS, Atom, feed or JSON Feed clue. Relative addresses are resolved by the browser. Duplicate targets collapse to one candidate, with the typed declaration preferred when the same URL also appears in the page.

It does not walk internal links, inspect pagination, guess CMS endpoints, fetch the feed or send the page to CreatorValet. Only the matched link’s short label accompanies its URL; unrelated paragraphs, forms, cookies, screenshots and resource bytes stay in the source tab. The panel repeats that payload before the user sends it, and downloaded JSON is the fallback when direct tab transfer is blocked.

RSS, Atom and JSON Feed are identified from explicit source

application/rss+xml and application/rdf+xml are reported as RSS, application/atom+xml as Atom, and application/feed+json as JSON Feed. RSS 2.0 is an XML syndication format with a channel and items, as described by the current RSS 2.0 specification. Atom has its own IETF specification, while JSON Feed defines a JSON object with a version URL and items in its version 1.1 specification.

A normal visible link has no authoritative response type before it is requested. Its path and label can therefore narrow the search but cannot establish the format. The card keeps “format not declared” when the clue is generic instead of guessing from branding or the site’s CMS.

Why this finder does not accept a remote website URL

The leading web tools accept a URL, fetch the page, probe common endpoints and preview the first feed items. That workflow needs a server-side arbitrary-URL fetcher. It also creates an SSRF boundary: every DNS answer and redirect must remain public, response bytes and time need hard limits, and the actual connection must be pinned to the address that passed the check while preserving the requested Host name and TLS identity.

CreatorValet’s current static Cloudflare runtime cannot provide that last guarantee for an arbitrary external origin. A preflight DNS lookup and the later platform fetch can resolve independently, leaving a rebinding window. The product therefore does not hide a server proxy behind the form, and it does not present pasted-URL behavior that cannot be operated safely. Inspecting the user’s already-open page is the bounded alternative.

How to verify a candidate after discovery

Open the candidate deliberately and check that the response is actually a feed rather than a branded error page or HTML document. Then use the W3C Feed Validation Service for RSS or Atom syntax. A successful syntax check still says nothing about freshness, complete article content, image quality, stable identifiers or whether a reader can access the URL over time; those are separate operational checks.

CSV export preserves URL, declared format, evidence, MIME, matched label and the literal not-fetched network status. JSON keeps the complete source receipt. If your adjacent job is collecting every ordinary hyperlink rather than just feed clues, the URL extractor uses the same current-page boundary and gives you deduplication, sorting and domain-only output.