HTTP asks every proxy that forwards a message to add itself to the Via header, defined in RFC 9110. Each hop appends an entry of the protocol version and a name, which can be a hostname or a pseudonym:
Via: 1.1 proxy.example.net (squid/6.10), 1.1 cache-fra
Read left to right, it traces the route the message took. Proxies use it to detect forwarding loops; administrators use it to see which caches a request passed.
Why it matters for proxy buyers
A Via header on a request tells the website that a proxy is involved. That is the difference between an anonymous proxy, which hides your IP but adds Via, and an elite proxy, which adds nothing. Commercial scraping proxies normally omit it.
Check your proxy
curl -s -x http://USERNAME:PASSWORD@HOST:PORT http://httpbin.org/headers
Look for Via in the echoed headers. Test against an http:// URL: HTTPS requests pass through a proxy inside an encrypted tunnel, so the proxy cannot add headers to them and the test would always come back clean.
Common confusion
You will often see Via on responses from the sites you scrape, such as via: 1.1 varnish or a CloudFront entry. That header was added by the site's own CDN or cache, a reverse proxy, on the way back to you. It says nothing about your proxy, and nothing about whether the site spotted you. Its sibling on requests is X-Forwarded-For, which carries addresses rather than proxy names.