SEO Tools
HTTP Link header parser.
Paste a Link: response header. See every link split out with its rel values, hreflang, type, media, anchor, and decoded title*. Missing rel, duplicate parameters, and relative URIs are flagged.
Header
Privacy note: parsing runs in this browser. Nothing is uploaded and no URL you paste is contacted. This checks header syntax only; it does not confirm the linked pages exist.
Parsed links
Paste a header and press Parse header.
What this parser does
Grammar
Follows RFC 8288 section 3: <URI-Reference> then ;-separated parameters, each a token or quoted-string. Commas and semicolons inside quotes do not split anything.
Parameters
Lists each rel (space-separated values become separate relations), plus anchor, hreflang, type, media, and title. title* is decoded as an RFC 8187 value (UTF-8 or ISO-8859-1) and wins over title.
Diagnostics
Flags a missing rel, repeated rel/type/media/title, deprecated rev, malformed quoting, and relative URIs. Enter a base URL to see them resolved.
Link header questions
How is this different from the hreflang checker?
The hreflang checker reads language alternates in HTML, headers, and sitemaps. This tool parses any Link header, whatever the relation: canonical, preload, next, prev, alternate, or extension types.
What happens when a parameter is repeated?
RFC 8288 says rel, media, title, title*, and type must not appear more than once in a link and later occurrences are ignored. hreflang may repeat. The parser reports the repeat and uses the first value.
Does a clean result mean the link works?
No. It confirms the header is well-formed. It does not fetch the target or confirm that a browser or search engine will act on it.
Can one Link header carry several links?
Yes. Separate the link-values with commas, as in a header that lists both an alternate language page and a canonical URL. MDN documents specifying multiple links in one header, and RFC 8288 defines the field as a comma-separated list. Commas inside quoted strings are not separators.
Does Google read canonical and hreflang from a Link header?
Yes, for web search. Google documents a rel="canonical" HTTP header as a method, including for non-HTML documents such as PDF files, and documents HTTP headers as one of three equivalent ways to give language and region alternates, again noting they suit non-HTML files.
Which relations work reliably for performance hints?
MDN says the only relations that work reliably in a Link header are preconnect and preload; stylesheet is reported to work reliably only in Firefox. Chrome's documentation also describes using the Link header with preload or preconnect on the main response when your server can respond right away.
How do I see the Link header a URL returns?
Open your browser's developer tools, load the page, and look at the response headers in the Network panel, or run curl -I on the URL from a terminal. Copy the value after Link: into this parser. Some servers treat HEAD differently from GET, so confirm with the browser if the result looks wrong.
Sources and standards
- RFC 8288: Web LinkingLink header ABNF, rel, anchor, and target attributes (October 2017).
- RFC 8187: Indicating Character Encoding and Language for HTTP Header Field ParametersThe title* ext-value encoding.
- RFC 3986 section 5Relative reference resolution.
- MDN: Link headerSame semantics as the HTML link element; multiple links; reliable relations. Retrieved 2026-10.
- Google Search Central: consolidate duplicate URLsThe rel="canonical" HTTP header, including for non-HTML documents such as PDFs. Retrieved 2026-10.
- Google Search Central: localized versions of your pagesHTTP headers as a way to indicate language and region alternates. Retrieved 2026-10.
- Chrome for Developers: Early HintsUsing the Link header with preload and preconnect on the main response. Retrieved 2026-10.
What a Link header is
A Link header is an HTTP response header that does what an HTML <link> element does, but without needing an HTML document. RFC 8288, "Web Linking", defines the model (a link has a target, a relation type and optional attributes) and its serialisation in the Link header field. MDN notes that the Link header has the same semantics as the HTML <link> element, and that one benefit of using the header is that the browser can start preconnecting or preloading resources before the HTML body has been processed. It is also the only option for files that have no <head> at all, such as PDFs and images.
Reading one, piece by piece
Take a header with two links:
Link: <https://example.com/fr/>; rel="alternate"; hreflang="fr", <https://example.com/>; rel="canonical"
- The comma separates link-values, so there are two links. Commas and semicolons inside quoted strings do not split anything.
- Each link starts with the target in angle brackets. Here the first target is the French page.
- Parameters follow, separated by semicolons. The first link has relation
alternateandhreflangoffr; the second is thecanonicalURL.
The parser shows each link in a row with its resolved target, its list of relation types, and the other attributes it found. A rel value containing several space-separated words, such as rel="alternate stylesheet", is shown as separate relations. If the target is relative (for example </fr/>), enter a base URL and the parser resolves it using the reference-resolution rules of RFC 3986, section 5.
Relations you will meet in practice
| rel | Typical use in a Link header |
|---|---|
canonical | Google documents a rel="canonical" HTTP header as a way to indicate the canonical URL for documents it supports, including non-HTML files such as PDFs, and notes the method applies to web search results only. |
alternate with hreflang | Google documents HTTP headers as one of three equivalent ways to indicate language and region versions, and describes the header as especially useful for non-HTML files such as PDFs. |
preload, preconnect | MDN says these are the only relations that work reliably in the Link header across browsers, letting the browser fetch or connect early. |
stylesheet | MDN states it does not work reliably across browsers in a Link header (only in Firefox), so use an HTML link for stylesheets. |
next, prev, help and others | Registered relation types that describe how resources connect; the parser lists them but does not judge whether a browser or search engine acts on them. |
Worked examples
A PDF with a canonical page
A PDF download that duplicates an HTML article can return Link: <https://www.example.com/guide/>; rel="canonical" in its response. Paste that header: the parser shows one link, relation canonical, target https://www.example.com/guide/, and no warnings. Remove the rel parameter and it reports a missing rel, because RFC 8288 says the rel parameter must be present in a Link header field.
A repeated parameter
For <https://example.com/a>; rel="next"; rel="prev" the parser reports that rel is repeated and uses the first value, next. RFC 8288 says rel must not appear more than once and that occurrences after the first are ignored, so what you meant by the second one is silently lost in the wild.
A non-ASCII title
The title* parameter uses the RFC 8187 encoding, for example title*=UTF-8''Caf%C3%A9. The parser decodes it to "Café" and, as the RFC specifies, prefers it over a plain title on the same link.
Common mistakes
- Leaving out the angle brackets around the target, or putting them around the whole header.
- Unquoted values with delimiters. A value containing a comma or semicolon must be quoted or it will be split in the wrong place.
- Sending the header only on some responses. A canonical or hreflang header must be on the response for the URL it describes, so check the header on that URL with
curl -Ior your browser's network panel. Some servers answer HEAD requests differently from GET, so confirm with a real GET too. - Conflicting signals. If the HTML says one canonical and the header says another, you have asked search engines two different questions; make them agree.
- Assuming a clean result is a working result. The parser checks grammar. It does not fetch the target, so a perfectly formed header can still point at a 404.
Who this is for, and what it does not cover
It is for developers and SEO specialists debugging headers from a CDN, a proxy or an application server, or writing them for the first time. It does not send requests, check status codes or follow redirects. To check language alternates in HTML, headers and sitemaps together, use the Hreflang Checker; for the canonical tag in HTML, the Canonical Tag Checker.