3 ms·
No, it's not. As a previous comment had pointed out before as well: You wait for handshakes, fake a deauth packet of the handshaking client, spoof an access p
by y4mi 7y ago
No, it's not.
As a previous comment had pointed out before as well:
You wait for handshakes, fake a deauth packet of the handshaking client, spoof an access point with the same SID and wait for the deauth'd client to try a reconnect.
Voila, cleartext PSK without any bruteforcing.
And it's not solveable either. You can't use fingerprinting as this would make mesh lans and quick access point failover impossible.
- mlyle 7y ago> spoof an access point with the same SID Pwnagotchi does not do this, despite your errant assertions. E.g. source code: https://github.com/evilsocket/pwnagotchi/blob/64e677f5df8f9bffd6fc56b71c94cbddc07c13e1/bin/pwnagotchi#L92 https://github.com/evilsocket/pwnagotchi/blob/64e677f5df8f9b... > Voila, cleartext PSK without any bruteforcing. Pretending to be an access point and going through a handshake doesn't let you retrieve the pre-shared key. (Unless the client is vulnerable to downgrade attacks-- which hasn't been a big consideration in more than a decade). Evil twin attacks are powerful but don't achieve what you say. The station sends to the access point a message authentication code based on nonces and the pairwise master key, which in turn is based on the "network password". It's produced using a series of HMACs and isn't an operation that can be inverted without brute force. https://en.wikipedia.org/wiki/IEEE_802.11i-2004#/media/File:4-way-handshake.svg https://en.wikipedia.org/wiki/IEEE_802.11i-2004#/media/File:...