On 6/3/25 13:09, Alan Coopersmith wrote:
> [I'm not sure how the attacker is supposed to get the victim to make a
>   requests call using a URL the attacker controls, but that didn't stop
>   them from getting a CVE issued for this. -alan- ]

Suppose that a server (like a web scraper) receives URLs that are
attacker-controlled, validates that the point to the expected domain
name, and then fetches them.  In this case, Requests will send
credentials for a domain name that is *not* the one that it is
supposed to send them for, which is clearly a vulnerability.

It's definitely better to reconstruct the URL from scheme, authority,
path, and query before sending the request, but I am almost certain
there are servers in the wild that do not do this.
-- 
Sincerely,
Demi Marie Obenour (she/her/hers)

Attachment: OpenPGP_0xB288B55FFF9C22C1.asc
Description: OpenPGP public key

Attachment: OpenPGP_signature.asc
Description: OpenPGP digital signature

Reply via email to