// admin tools · 25
User-agent parser
Break any user-agent string into browser, engine, OS and device, spot crawlers — search, AI training, link previews, SEO — and in-app browsers, or paste a whole access log for counts.
// runs locally — nothing is uploaded
client hints · this browser only
How it works
What it does
This parser reads a User-Agent header and tells you the browser and version, rendering engine (Blink, Gecko, WebKit, Trident or EdgeHTML), operating system — with Windows NT numbers turned into marketing names — and the device type, vendor and model where the string reveals them. It recognises more than 90 crawlers and HTTP clients, labels each by purpose, and flags in-app browsers such as Instagram, Facebook and TikTok. It is written from scratch and runs entirely in your browser; nothing you paste is sent anywhere.
How to use it
- The page opens with your own UA. Paste any other, or try an example.
- --bulk takes raw access-log lines or bare UAs, one per line, and counts browsers, operating systems, device types, bots and bot purposes. Copy the result as CSV.
- --bots lists every crawler the parser knows, with the token to use in
robots.txt. - Under client hints, Chromium browsers show what
navigator.userAgentDatareports — often more accurate than the UA string.
Why user-agents lie
Every modern browser claims to be Mozilla/5.0, Chrome mentions Safari, and Edge mentions Chrome. Browsers also freeze details for privacy: Chrome's reduced UA zeroes the minor version and, on Android, reports Android 10; K instead of the real version and model. Safari and Chrome on a Mac report macOS 10.15.7 whatever the actual release, and Windows 11 still says Windows NT 10.0. iPads request desktop sites by default and send a Mac UA. Treat a UA as a hint, not proof.
Bot purposes
- search
- Indexes pages for a search engine — Googlebot, Bingbot, Applebot, DuckDuckBot.
- ai-training
- Collects content to train models — GPTBot, ClaudeBot, CCBot, Bytespider.
- ai-search / assistant
- Fetches pages for AI answers or because a user asked — OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-User.
- preview
- Builds link cards when a URL is shared — facebookexternalhit, Twitterbot, LinkedInBot, Slackbot.
- seo
- Commercial backlink and audit crawlers — AhrefsBot, SemrushBot, MJ12bot.
FAQ
Can I trust a request that says it is Googlebot?
No. Anyone can send that string. Verify with a reverse DNS lookup that resolves to googlebot.com or google.com and a matching forward lookup, or check Google's published IP ranges. The same applies to other major crawlers.
Does blocking GPTBot or ClaudeBot hurt my search ranking?
No. They are separate from search crawlers. Disallowing them in robots.txt leaves Googlebot and Bingbot untouched. Google's training opt-out is the separate Google-Extended token.
How do I tell Windows 10 from Windows 11?
Not from the UA string. In Chromium browsers, request the platformVersion client hint: 13 or higher means Windows 11.
Is my user-agent sent to you?
No. Parsing happens in this page's JavaScript and there are no network requests.