User Agent Parser

How it works

1. 01

   Paste a User-Agent string into the field, or press "Use my browser’s" to load the one this browser is sending right now.

2. 02

   An ordered table of token rules is tested against the string. The first rule that matches wins, and the rules run most-specific first, because nearly every browser prints the tokens of the browsers it was compatible with.

3. 03

   Four answers come out: browser and version, rendering engine and version, operating system and version, and device type. A bot rule table runs first and reports separately.

4. 04

   The matched token is shown beside each answer, so you can check the reasoning rather than trust it. Copy the one-line summary if you need it in a ticket.

Why the rule order is the whole algorithm

A User-Agent is a compatibility ledger, not an identifier. Edge prints Chrome and Safari; Chrome prints Safari; Amazon Silk prints "like Chrome/"; Headless Chrome prints the substring "Chrome/" inside its own token. So a parser that tests for Chrome first calls all four of them Chrome. The table is ordered from the narrowest token to the widest, and the first match wins — which means adding a rule in the wrong position silently rewrites the answer for every browser below it.

Edg/ → OPR/ → SamsungBrowser/ → Silk/ → HeadlessChrome/ → CriOS/ → Chromium/ → Chrome/ → Firefox/ → Version/+Safari/ → Safari/

Use cases

Questions

Can a User-Agent be trusted?

No. It is a request header the client chooses, and every browser has a setting or an extension that changes it. Treat the answer as what the request claims to be. It is fine for logs, analytics and bug reports; it is not an identity check and it is not a security control.

Why does Edge say it is Chrome and Safari?

History. Every browser added the previous market leader’s tokens so that servers sniffing for it would serve the good page — which is why almost every string still starts with Mozilla/5.0. Edge prints Chrome and Safari and then Edg/ at the end, so the only correct way to read it is narrowest token first.

Why is my browser reported as Chrome or Safari?

Some browsers deliberately do not identify themselves. Brave has shipped no Brave token since 2016 and reports as Chrome; Tor Browser reports a fixed generic Firefox string; several privacy modes do the same. There is nothing in the string to detect, so this tool does not guess.

Why does Windows always come back as 10 or 11, and macOS as 10.15.7?

Both are frozen. Windows 11 reports NT 10.0 with no separate token, so the two cannot be told apart from the UA. Safari has reported Mac OS X 10_15_7 since Big Sur regardless of the real version. Sec-CH-UA-Platform-Version carries the real number if the server asks for it.

Can it tell an iPad from a Mac?

Only when the iPad sends an iPad token. In desktop-class mode iPadOS sends a Macintosh string that is byte-for-byte a Mac’s, and no parser can undo that. The usual workaround is checking navigator.maxTouchPoints in the page, which this tool does not do because it parses a string you paste rather than probing your device.

Should I use this to decide which code to run?

No. Feature-detect instead — test for the API you are about to call. UA sniffing breaks on the next release, on every fork, and on anything spoofing. Parsing a UA is for reading logs after the fact, which is a different job from branching in production code.

Is anything sent to a server?

No. The rule tables are a few dozen regular expressions and all of them run in the page. The string is not uploaded, logged, or stored between visits, and pressing the button to load your own User-Agent reads it locally and does nothing else with it.