URI schemes: the full list, and which six are special

What is a URI scheme?

A URI scheme is the part before the colon: the https in https://example.com, the mailto in mailto:someone@example.com, the spotify in a link that opens an app. It tells whatever receives the URL what kind of thing it is looking at, and IANA keeps the register of them — 426 at the time this page was generated, each marked Permanent, Provisional or Historical.

Almost every list of URL schemes on the web is a copy of that register with the status column quietly dropped. This one keeps it, and adds the column no copy has, because the register turns out not to answer the question most people arrive with.

That question is how a browser will actually treat the scheme. The answer lives in a different specification entirely, and it takes the form of a table with six rows.

How to use it

1. Type a scheme into the box. Anything at all, registered or invented. Five inputs are handed to your browser's own URL parser and the answers appear underneath.
2. Read which lines say they differ. Each check is compared against an ordinary registered scheme, so a line marked as differing is a special-scheme rule being applied and nothing else.
3. Search the register below. By scheme name or by description. Each row carries its IANA status, whether the URL Standard treats it specially, and its default port if it has one.

Six special schemes, and the register cannot tell you which

The WHATWG URL Standard — the specification browsers actually implement — defines a special scheme as one listed in a table it prints in full. The table has six rows: ftp with default port 21, file with no port at all, http with 80, https with 443, ws with 80 and wss with 443. A URL is special if its scheme is one of those six, and every difference on this page follows from that single predicate.

Line the two sources up and they do not correspond. All six special schemes are registered Permanent, which sounds like a rule until you notice that 93 other schemes are also Permanent and are parsed as ordinary ones. gopher is Permanent, with its own RFC, and your browser gives it no special treatment whatsoever. mailto is Permanent and ordinary. Meanwhile a scheme you invent this afternoon, which is in no register at all, is parsed exactly like gopher.

So registration status tells you whether a scheme has been through a standards process. It tells you nothing about parsing. The two facts are maintained by different bodies for different purposes, and conflating them is the most common mistake in this corner of the web.

What being special actually changes

The default port is dropped. Parse https://example.com:443/ and the port disappears, because 443 is the default for https; parse git://example.com:443/ and the port stays, because git has no default to compare against.

Backslashes become forward slashes. The string https:\\example.com\path parses as https://example.com/path — the parser silently repairs it. Under an ordinary scheme every backslash survives untouched.

An empty host is refused, and what happens instead is the surprising part. A special scheme may not have an empty host, so https:///path does not fail: the first path segment is promoted into the host, and you get a URL whose host is path. Under an ordinary scheme the empty host is preserved and the path stays where you put it. file is the exception among the six — it is the one special scheme allowed an empty host, which is why file:///etc/hosts works the way it does.

The host is normalised. EXAMPLE.München.DE becomes example.xn--mnchen-3ya.de: lowercased and run through IDNA. An ordinary scheme keeps the case and percent-encodes the non-ASCII characters instead.

And a special scheme demands a host. new URL("https:") throws a TypeError, while new URL("git:") is a perfectly valid URL that parses to git: and nothing else.

One difference that is not real

Dot-segment removal is regularly listed as special-scheme behaviour, and it is not. A path of /a/./b/../c collapses to /a/c under https, under git, and under a scheme nobody has ever registered. Any URL with an authority gets its dot segments removed, so this is a property of having a host, not of being special.

We include this because the claim was in our own notes before the page was written, and checking it against the parser is what removed it. The checks above are run in your browser rather than quoted from a specification for the same reason: a page about what browsers do should ask one.

Two honest limits. The register is a snapshot, dated in the panel below the table, and schemes are added continuously — a scheme registered last week may not be here, though it will parse identically to the ordinary ones that are. And a scheme being registered says nothing about whether anything on your machine will open it: that is a matter of what software you have installed, which no web page can inspect.

Why is it free?

The register travels with the page as a few kilobytes and the parsing checks run in your own browser. Nothing you type is uploaded, nothing is logged, and there is no account to create.

No sign-up, no limits, and no watermark on anything you copy.