Restrict CONNECT ports rule, GET requests and a RAT

[wyrm]

Yesterday, we detected a RAT on one of our endpoints that was connected to a C2 server (all undetectable by McAfee).  The device was connected to 1 IP over port 80 and was sending GET requests to another IP over port 43434.  Since the default port blocking rule in the MWGs is only for CONNECT events, the GET requests were allowed on 43434.

Does anyone know if there is any risk to adding GET commands to the default Restrict CONNECT Ports policy?  Or should I be looking at blocking all traffic where URL.port is not in the Allow CONNECT Ports lists, regardless of the HTTP command?  How would you suggest I handle this?

[smasnizk]

Hi Wyrm,

this idea won't work as simple it might sounds. The Rule Set you're talking about is restricted to "CONNECT" only. If you want to build the same for GET request as well, you can use same criteria but be careful of other services your company might use. You will simply build this over time by allowing some which might be crucial. 

If you do have a real world sample including connection traces you can build a rule set for community, other customer could also benefit from it.

Regards,

Sergej

[wyrm · Accepted solution]

All I did to resolve this was take the existing Restrict CONNECT Ports rule and removed the criteria where "command.name equals "CONNECT."  Simple fix that now blocks all non-approved ports.  I'm not sure why this wasn't default.