The vulnerability is a classic LDAP injection in the proxyauth addon of mitmproxy. The analysis of the patch commit 71c9234057922bc29b9734ec408d712113d294d2 shows that the Ldap class's __call__ method was modified. Prior to the patch, the username from the client was directly used to format an LDAP search filter string. This allowed an attacker to inject special LDAP filter characters (like *, (, )) to alter the logic of the LDAP query, leading to an authentication bypass. The fix involves adding a new make_search_filter method that sanitizes the username by escaping these special characters using the ldap3.utils.conv.escape_filter_chars function before constructing the filter. The Ldap.__call__ function is the vulnerable function as it's the one that takes the untrusted input and uses it in a dangerous way.