The DoD’s 6,000-plus person Cyber Mission Force (CMF) needs a way to deploy realistic, automated cyber adversaries on its training platform to develop and practice the skills required to execute defensive cyber operations missions

Background

The solution should enable training planners to conveniently build an offensive campaign with automated offensive actors, enable more realistic training of defensive teams by emulating the actions and decisions of real, advanced persistent threat (APT) groups, and reduce the manpower costs of having highly-trained, high demand human operators act as adversaries for training scenarios.

The solution should:

allow training planners to create their own attack campaigns for training either from components (reconnaissance, exploitation, lateral movement, etc.) or from a library of known tactics used by real-life threat actors, based on open source cyber threat intelligence. The solution should
allow for planners to control characteristics of the automated adversaries, such as objectives, level of stealth, etc. and/or have adaptive agents that can learn from their environment and make autonomous decisions based on goal criteria.
have an intuitive graphical user interface (GUI) to allow planners who do not possess first-hand operational tradecraft knowledge to effectively manage adversary behavior and to audit what actions the automated adversaries are taking.

Preference will be given to proposals that demonstrate product maturity and deployment validation. However, partial solutions that uniquely address specific requirements will be considered, as well as teaming arrangements that combine multiple capabilities to achieve the desired solution. Solutions would need to be capable of deployment on a closed / air-gapped network.