RSS & Atom Feed Validator

About the RSS & Atom Feed Validator

A feed that a reader cannot parse simply stops updating, and nobody tells you. Enter a feed address, paste the feed or open the file, and the validator checks it the way feed readers read it: XML that strict parsers reject (an unescaped &, an  , a broken tag), what the specification requires (title, link and description, Atom’s id and updated, JSON Feed’s version and items), dates in the right format, guids that are unique and honest about being permalinks, enclosures with a URL, length and type, and the problems that confuse readers in practice — relative links in content, several enclosures, items dated in the future.

It reads RSS 2.0 (and the older 0.9x), RSS 1.0 (RDF), Atom 1.0 and JSON Feed 1.0 and 1.1. Each problem comes with its line, a concrete fix and the rule behind it; repeated item problems are grouped with examples. You also get a reader-style preview of the feed (as plain text: nothing from the feed is run or loaded) and the same feed converted to JSON Feed 1.1. For an address, the server’s answer is checked too: status, redirects, content type, caching headers and compression.

Common uses

• Finding out why a blog, newsletter archive or podcast stopped updating in feed readers and aggregators.
• Checking a hand-written or generated feed before publishing it.
• Checking that a site’s pages advertise their feeds with <link rel="alternate">, so readers and browsers can find them.
• Converting an RSS or Atom feed to JSON Feed, for apps that read JSON.

What is checked

• XML: well-formedness with line and column (unescaped & and <, undefined entities such as  , mismatched or unclosed tags, control characters, a second root, a late XML declaration), the declared encoding against the file, and undeclared namespace prefixes.
• RSS 2.0: version="2.0", one <channel> with title, link and description; atom:link rel="self"; language codes; RFC 822 dates (with the corrected form when it can be worked out); ttl, the channel image (width at most 144, height at most 400), textInput, cloud, skipHours and skipDays; every item’s title or description, link, guid (unique, and isPermaLink="false" when it is not an address), pubDate, enclosures (url, length in bytes, MIME type; one per item), author as an email address, relative addresses in content, and elements that are not part of RSS.
• Atom 1.0: exactly one id, title and updated in the feed and every entry, ids that are absolute IRIs and unique, RFC 3339 dates, authors, rel="self" and alternate links, content or an alternate link in every entry, the summary RFC 4287 requires for src or binary content, and the XHTML div wrapper.
• JSON Feed: syntax with line and column, version, title and items, home_page_url and feed_url, item ids (present, strings, unique), content_html or content_text, RFC 3339 dates, attachments, deprecated author, and keys that are not part of the format (extensions must start with _).
• The server’s answer (URL mode): status, temporary or permanent redirects, Content-Type, ETag / Last-Modified (so readers can ask “has it changed?”), compression and the 2 MiB size limit.

Sources

The checks follow the RSS 2.0 specification and the RSS Advisory Board’s RSS Best Practices Profile, RFC 4287 (Atom), JSON Feed 1.1 and the RSS 1.0 specification. Dates follow RFC 822 for RSS and RFC 3339 for Atom and JSON Feed.

Limitations

• It checks the feed file, not the links in it: it does not fetch item pages, images or enclosures. The Broken Link Checker checks links on a page.
• Podcast-specific rules of Apple Podcasts, Spotify or the Podcasting 2.0 namespace (artwork size, categories, explicit flags) are not checked; their namespaced elements are accepted as extensions.
• Fetched feeds are read as UTF-8. A feed in another encoding is flagged, and accented characters may show as � in the preview; open the file instead to have it decoded by its XML declaration.
• Content in items is shown as plain text in the preview: images, video and formatting are not rendered, and nothing from the feed runs.
• A site can answer MySmartCoPilot’s server differently from a feed reader, and the server reads at most 2 MiB. Each visitor can run a limited number of fetches per hour; pasted and opened feeds have no limit.

Privacy

In Feed URL mode the address is sent to MySmartCoPilot’s server, which fetches the feed. MySmartCoPilot does not store the address, the feed or the result, and the server log records only the host name, status code and time of each request, never the full URL or your IP address. Pasted text and opened files stay in your browser.