User Agent Parser

A user agent string is a short line of text every browser and many other HTTP clients (bots, apps, command-line tools) send with essentially every request, identifying what software is making the request — historically the browser name and version, the rendering engine, and the operating system. Servers and analytics tools read it to decide how to respond: serving a mobile-optimized layout, blocking a known bad bot, or simply logging what browsers real visitors use.

This is a direct result of decades of browser vendors copying each other's identifying tokens to avoid being blocked by sites that only checked for one specific browser — a modern Chrome user agent literally contains the substring 'Safari' (because early sites checked for it), and Edge's user agent contains both 'Chrome' and 'Safari' tokens alongside its own 'Edg/' identifier, because Edge is built on the same Chromium engine Chrome uses. Correctly parsing a user agent therefore requires checking the most specific, most recently-added tokens first (like 'Edg/' or 'OPR/') before falling back to more generic ones (like 'Chrome/' or 'Safari/') — exactly the order this tool's parser follows.

The rendering engine (also called the layout engine) is the underlying software that actually parses HTML/CSS and draws the page — Blink, Gecko, WebKit, and the older Trident are the major ones — while the browser is the actual application built around that engine, with its own UI, features, and branding. Multiple browsers frequently share the same engine: Chrome, Edge, Opera, Samsung Internet, Brave, and most other Chromium-based browsers all use Blink, which is why they behave nearly identically for web compatibility purposes despite being different products.

The user agent string is simply text the browser chooses to send — nothing stops any browser, script, or command-line tool from sending a completely fabricated value, and mainstream browsers themselves increasingly limit or 'freeze' the detail they expose in it for privacy reasons (Chrome's User-Agent Reduction initiative, for example, now often reports a generic recent Chrome version and OS bucket rather than exact minor versions). For genuinely reliable feature detection, checking whether a specific JavaScript API or CSS feature actually exists is far more robust than parsing the user agent and guessing — user agent parsing is best used for analytics, logging, and rough categorization, not as a security or compatibility gate.

This tool checks the user agent against a list of known signatures used by major legitimate crawlers (Googlebot, Bingbot, GPTBot, and dozens of others) as well as common command-line and scripting tools (curl, Wget, Python Requests), plus generic keywords like 'bot,' 'spider,' and 'crawl.' It can only recognize a bot that honestly identifies itself this way — a scraper deliberately impersonating a real browser's user agent (a very common and trivial evasion technique) will simply be classified as that browser, which is exactly why serious bot-detection systems combine many additional signals (request patterns, IP reputation, JavaScript execution, TLS fingerprinting) rather than relying on the user agent string alone.

Device-type classification from a user agent relies on specific tokens — 'Mobile' generally appears in phone user agents but is deliberately omitted from tablet user agents on some platforms (a stock Android tablet's user agent often omits 'Mobile' specifically so sites serve it the desktop layout, since tablets have enough screen space) — and some browsers let a user manually request 'Desktop site' mode, which changes the user agent sent for that session to look like a desktop browser entirely, overriding what the underlying device actually is.

It means the pasted string didn't match any of the recognized browser or OS signatures this tool checks for — this happens with genuinely unusual or custom user agents (an uncommon embedded browser, a heavily modified string, or one that's simply malformed or incomplete), or occasionally with a brand-new browser release using a token format not yet accounted for. Try comparing it against one of the library examples to spot what's different, or check the raw string for the tokens this tool's FAQ above describes.

No — your browser's user agent is read directly from JavaScript's navigator.userAgent property and parsed entirely in your browser; nothing is sent to a server, logged, or stored. The same is true for any custom string you paste in to parse manually.