FireEye’s Innovation and Custom Engineering (ICE) team released a tool called GoCrack that allows red teams to efficiently manage password cracking tasks across multiple GPU servers by providing an easy-to-use, web-based real-time UI to create, view, and manage tasks. Simply deploy a GoCrack server along with a worker on every GPU/CPU capable machine and the system will automatically distribute tasks across those GPU/CPU machines.
Password cracking tools are an effective way for security professionals to test password effectiveness, develop improved methods to securely store passwords, and audit current password requirements. Some use cases for a password cracking tool can include cracking passwords on exfil archives, auditing password requirements in internal tools, and offensive/defensive operations. GoCrack is another tool for distributed teams to have in their arsenal for managing password cracking and recovery tasks.
Keeping in mind the sensitivity of passwords, GoCrack includes an entitlement-based system that prevents users from accessing task data unless they are the original creator or they grant additional users to the task. Modifications to a task, viewing of cracked passwords, downloading a task file, and other sensitive actions are logged and available for auditing by administrators. Engine files (files used by the cracking engine) such as Dictionaries, Mangling Rules, etc. can be uploaded as “Shared”, which allows other users to use them in task yet do not grant them the ability to download or edit. This allows for sensitive dictionaries to be used without enabling their contents to be viewed.
GoCrack is shipping with support for hashcat v3.6+, requires no external database server (via a flat file), and includes support for both LDAP and database backed authentication. In the future, team plans on adding support for MySQL and Postgres database engines for larger deployments, ability to manage and edit files in the UI, automatic task expiration, and greater configuration of the hashcat engine. The server component can run on any Linux server with Docker installed. Users with NVIDIA GPUs can use NVIDIA Docker to run the worker in a container with full access to the GPUs.