Re: Suppressing/modifying some browser fingerprints (ie. Am I Unique?)
Lee <[email protected]> Fri, 17 Jul 2020 19:54:17 -0400
| Newsgroups | gmane.comp.web.privoxy.user |
|---|---|
| Message-ID | <CAD8GWssQY=2S3PWOExjTX1qRRr3o+bdTJjQ--DW7ekBvFeP4Hg@mail.gmail.com> |
On 7/17/20, Nicholas Bastin <[email protected]> wrote: > On Fri, Jul 17, 2020 at 7:20 PM Lee <[email protected]> wrote: > >> So it looks like the diagram under 'how it works' at >> http://www.proxfilter.net/proxhttpsproxy/ >> right? >> >> I've done "bump in the wire" stuff before and never really liked it. >> >> The big thing I like about doing everything in Privoxy is that I get >> to check the cert in the browser. Any site where I care about >> security should show up as having a legit cert; the sites where I'm >> doing https inspection show up with a cert from "Billy Bob's Beer, >> Bait and CA Store". > > All bumping issues new certificates from your roots, which you of course > put in your host certificate store, just like you would doing it inside > privoxy. Your egress bump handler should be validating the actual server > certificates against your policy (e.g. via an SSL observatory, local hash > for monitoring deltas, etc.). That's a problem right there.. my "policy" has been "wherever the software does". I don't really know how firefox/curl/wget validates certs :( > The difference with handling it in a > pipeline is that you break out the pieces so they can be modified > independently, meaning you can shift protocols or cipher suites in your > bump handler without privoxy having to know anything about it, and let > privoxy do what it does best. Yes, I see the attraction. But me being able to see which certs are used for what __in the browser__ is much more attractive to me. hrmm.. altho it's probably more that I'm not all that confident of my ability to create and enforce a safe & secure policy for handling TLS and certificates, so I'm going for ease of validation.. Regards, Lee