User-agent parser: see what a server can infer from a UA string

A user agent is a self-declared string, not an identity. This tool parses a UA into the product token, browser, engine, operating system, device class, and a small set of known bots. It runs in your browser and does not call a lookup service.

How to use it

1. Paste a user-agent string, or choose an example.
2. Read the parsed fields.
3. Use the product token when writing a robots.txt group or a server-side rule.

What the parser detects

• Product token: the first token, such as Mozilla, Googlebot, or GPTBot.
• Browser: Edge, Chrome, Firefox, Safari, Opera, or an embedded WebKit client.
• Engine: Blink, Gecko, or WebKit.
• OS: Windows, macOS, iOS/iPadOS, Android, Linux, ChromeOS.
• Device class: bot/crawler, mobile, tablet, desktop.
• Known bot: a small list including Googlebot, Bingbot, GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, and CCBot.

Why it matters

A server uses the user agent for logging, analytics, and access policy. A crawler should use a descriptive token with a contact URL, not a browser impersonation. A site owner should not treat the user agent as authentication; it is trivially spoofed.

For AI crawler policy, the token is what robots.txt matches. The AI crawler policies page lists the main tokens and how to allow or block them.

Limits

• Parsing is heuristic. User-agent strings are inconsistent and some clients lie.
• The bot list is not exhaustive.
• The tool does not verify the client’s IP, TLS fingerprint, or behaviour.
• A UA string does not prove a client is a browser or a crawler.

Sources

• RFC 9110 — User-Agent
• MDN — User-Agent