User Agent Parser

A User-Agent parser reads the User-Agent request header a browser sends and breaks it into the browser name and version, the rendering engine, the operating system, and the kind of device the request came from. Every HTTP request carries this string, and it is the single most common way a server learns what is on the other end of a connection without asking. The header itself is defined in RFC 9110, the HTTP semantics specification, and its format is an unstructured list of product tokens and comments that has grown historically messy.

That messiness is the reason a parser exists. A modern Chrome string still begins with "Mozilla/5.0" and still contains the word "Safari", because browsers spent two decades copying each other tokens to pass server sniffing tests. Edge carries "Chrome", Opera carries "Chrome", and Safari on an iPhone reports its real version in a separate "Version/" token rather than the "Safari/" one. Reading a User-Agent by eye is therefore error-prone: the tokens that look most obvious are frequently the ones left in for backward compatibility rather than the ones that identify the actual client.

This tool applies the detection order those quirks demand. It tests for the browsers that embed another browser token first, so Edge is not misread as Chrome and a Chrome-on-iOS string is not misread as Safari. It maps the Windows NT kernel number to its marketing name, converts the underscore-separated iOS and macOS versions back to dotted form, and classifies the request as desktop, mobile, tablet, or bot. It also flags known crawler and automation signatures, which is the field most useful when you are reading a web-server access log rather than debugging a single browser.

A worked example shows what the user agent string parser returns. Click the Edge on Windows sample and the Parsed result lists Microsoft Edge 126.0.0.0, Blink 537.36, Windows 10 / 11, Desktop, x86-64 (64-bit) and Bot / crawler No, even though the string also contains Chrome and Safari tokens. Now paste the Googlebot smartphone string, which claims Chrome on Android 6.0.1: the user agent parser still reads the browser and OS, but Device type becomes Bot / Crawler and the bot flag says Yes. That split is what you want when a crawler borrows a phone browser's identity in your logs.

Everything runs in your browser. The string you paste is never uploaded, which matters because a real User-Agent from your own traffic can hint at the exact device and build a visitor is using. There is no signup and parsing makes no network request; only the Copy button records a usage count, without the string. You can paste production log lines without sending them anywhere.