MetricRaven

User agent parser

Paste a user agent string to see the browser, version, operating system, device and engine. If it's a bot, you'll see which one, who runs it and why it crawls, from 112 known crawlers including Googlebot, GPTBot and ClaudeBot.

Try an example

Paste a user agent above, or pick an example, to see what it says.

Parsing runs in your browser. Nothing you paste is sent anywhere.

What is a user agent?

A user agent is the program that makes a web request: a browser, an app, a script or a crawler. It names itself in the User-Agent header of every request. Per RFC 9110, servers use it to work around known limitations of a client and for analytics about browsers and operating systems. Read more in the user agent glossary entry.

Here is a desktop Chrome user agent, token by token:

User-Agent
Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/157.0.0.0 Safari/537.36
Mozilla/5.0
A compatibility token. Every major browser sends it; it says nothing about the browser.
(Windows NT 10.0; Win64; x64)
The platform. Chrome now sends one fixed value per platform, so Windows 10 and 11 look the same.
AppleWebKit/537.36 (KHTML, like Gecko)
Engine compatibility tokens. Chrome's Blink engine still reports them.
Chrome/157.0.0.0
The browser and its major version. The rest of the version is frozen at 0.0.0.
Safari/537.36
One more compatibility token, kept so sites that look for Safari keep working.

Chrome has sent a reduced user agent since version 107 on desktop and 110 on Android. The minor version is always 0.0.0, macOS is always 10_15_7 and Android is always Android 10; K, whatever the real device runs (Chromium's User-Agent Reduction). The precise details moved to client hints such as Sec-CH-UA. So a parser can tell you the browser and major version, but not the exact Android or macOS release.

Googlebot user agent

Googlebot crawls as a smartphone for most sites and as a desktop browser for the rest. Google publishes these strings, with W.X.Y.Z where the current Chrome version goes:

  • User-Agent
    Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/W.X.Y.Z Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)
  • User-Agent
    Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Googlebot/2.1; +http://www.google.com/bot.html) Chrome/W.X.Y.Z Safari/537.36
  • User-Agent
    Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)
  • User-Agent
    Googlebot/2.1 (+http://www.google.com/bot.html)

Scrapers copy this string to get past bot filters, so the user agent alone proves nothing. Google's verification guide gives two checks:

  1. Reverse DNS: host 66.249.66.1 should return a name on googlebot.com, google.com or googleusercontent.com. A forward lookup of that name must return the same IP.
  2. IP ranges: match the IP against the JSON lists Google publishes for its common crawlers, special crawlers and user-triggered fetchers.

Details, range files and robots.txt rules are on the Googlebot crawler page.

AI crawler user agents

AI companies run separate crawlers for training, for their search index and for fetching a page when a person asks about it. Blocking one doesn't block the others. The token is what to match in logs and to name in robots.txt.

TokenOperatorPurpose
GPTBotOpenAIAI training
OAI-SearchBotOpenAIAI search index
ChatGPT-UserOpenAIAI assistant (user asked)
ClaudeBotAnthropicAI training
Claude-SearchBotAnthropicAI search index
Claude-UserAnthropicAI assistant (user asked)
PerplexityBotPerplexityAI search index
Perplexity-UserPerplexityAI assistant (user asked)
Google-ExtendedGoogleAI training
Applebot-ExtendedAppleAI training
BytespiderByteDanceAI training
CCBotCommon CrawlAI training
meta-externalagentMetaAI training
AmazonbotAmazonAI training

All 112 crawlers, with full user agent strings and robots.txt snippets, are in the crawler directory.

How MetricRaven tells bots from visitors

The parser above runs the same rules MetricRaven's tracker uses on every hit:

  • Bot words

    An empty user agent, or one containing words such as bot, crawl, spider, headless, lighthouse, curl, wget or python-requests, is ignored.

  • Stale Chrome

    Desktop Chrome updates itself. A desktop Chrome 20 or more major versions behind the current release is a scraper reusing an old string, so it's ignored.

  • Client hints

    When the user agent says one Chrome version and the browser's own Sec-CH-UA header says another, the hit is ignored. With “Use my browser's user agent”, the parser checks your browser the same way.

  • Data-center networks

    Real browsers on cloud servers (headless scrapers) are excluded by network, not user agent. See Exclude data-center traffic.

  • Crawlers

    Most crawlers don't run JavaScript, so they are reported from your server and verified against their operators' IP ranges. See Bots & AI crawlers.

Keeping bots out matters for revenue numbers, not just traffic. A thousand scraper hits on your pricing page look like a thousand visitors who didn't buy, and your conversion rate and revenue per visitor drop for no real reason.

Questions

What is a user agent?
A user agent is the text a browser, app or bot sends in the User-Agent header of every HTTP request to say what software it is. Servers use it to work around browser quirks, to block or allow bots, and for analytics about browsers and operating systems.
What is Googlebot's user agent?
Googlebot Smartphone, which crawls most sites, sends "Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/157.0.0.0 Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)". Google keeps the Chrome version current. Because anyone can copy this string, confirm a visit with a reverse DNS lookup: the IP should resolve to googlebot.com, google.com or googleusercontent.com, and that host should resolve back to the same IP.
Can a user agent be faked?
Yes. It's a plain header the client chooses, so scrapers copy browser and crawler user agents all the time. To trust a crawler, check its IP against the ranges its operator publishes or use reverse DNS. To spot a fake browser, look for signals such as a desktop Chrome far older than the current release.
How do I find my own user agent?
Open this page: the parser loads your browser's user agent automatically. In code, read navigator.userAgent in the browser, or the User-Agent request header on your server.
Why do bots and other browsers say Mozilla and Safari?
For compatibility. Early sites served better pages to browsers that said Mozilla, so later browsers copied the tokens of the ones before them. That's why Chrome's user agent also mentions Mozilla, AppleWebKit, KHTML, Gecko and Safari, and why many crawlers copy a browser's string and add their own name.

Sources checked 2026-11-18.

Count people, not bots, and see which of them pay

MetricRaven runs these checks on every hit. Bots, scripts and stale browsers never reach your visitor counts, AI crawlers are tracked on their own from your server, and real visits are tied to the revenue they bring.

14-day free trial, no card required.