Adaptive game assistant

The adaptive game assistant is a Python program allowing easy deployment and playing of adaptive cybersecurity games.

When bundled with a compatible game (and a few config files), the assistant allows the user to play through the game independently, with no help needed from a teacher or a supervisor. The assistant automatically runs Vagrant commands to prepare levels of the game, checks flags when player completes levels and provides hints.

The assistant was originally made as a part of my bachelor's thesis for the university's KYPO Lab. It was later improved to be used in a university course.

The games the assistant can run must be made with Vagrant and Ansible (as other games at KYPO). The Ansible playbooks need to have all tasks tagged, and there need to be game-specific config files in the resources/ subfolder.

Usage

The adaptive game assistant is ran by running ./assistant.py or python assistant.py in the project folder.

Basic commands:

• (S)tart - starts the game from level 1.
• (N)ext - continues the game to the next level.
• (E)xit - properly ends the game and exits the assistant.
• (C)eck - checks versions of all required apps.
• (H)elp - displays a full list of commands.
• hin(T) - displays hints, offers new hints.
• (L)og - saves data from the game into a file.

The project's wiki has a user guide with examples: Assistant guide

Requirements

The assistant requires Python 3.7 or higher to run.

It also requires the PyYAML package of version 5.1 or higher. (You can install it with pip install -r python-requirements.txt)

The game the assistant sets up will require Vagrant and VirtualBox to be installed. Virtualbox 6.0 or higher and Vagrant 2.2.5 or higher are recommended. (The exact requirements of individual games may differ.)

The wiki has an in-depth installation guide: Installation

Python module structure

The main file, assistant.py makes use of the files from the adaptive_game_module/ folder, and transforms the user's commands into method calls on the objects.

The adaptive game module contains:

• adaptive_game.py. The Game class represents the state of the game.
• flag_checker.py. The FlagChecker class checks if flags are correct.
• hint_giver.py. The HintGiver class keeps tracks of taken hints and gives new ones.
• level_selector.py. The LevelSelector class helps decide which level to go to next.

Game requirements

Ansible tags

The assistant expects some things of the Ansible playbooks.

• All tasks should be tagged
• Tasks for game setup and the first level are tagged setup
• Tasks for level past the first are in the format level + [number] + [branch, optional] i.e. "level3", "level4a"

Game config files

Besides tagged ansible playbooks, each adaptive game needs a few config files to work. The config files are mostly YAML lists and dicts. assistant.py looks for config files in the resources/ folder.

The needed files are:

• levels.yml
• hints.yml
• level_keys.yml
• level_requirements.yml
• tools.yml

More about the config files and their format is on the wiki: Config files

I want to try out the assistant, but I don't have an adaptive game

The simplest way to try the assistant out with no access to another adaptive game is:

1. Download the thesis archive of my Adaptive Cybersecurity Games thesis.
2. In the archive, replace the assistant/ folder with the folder of this repository. (The assistant included with the thesis is an earlier version.)
3. Extract game.zip from the archive. (So that you have a game/ folder.)
4. From the file game/provisioning/roles/attacker/tasks/main.yml, delete lines 34-39.
5. Run the assistant.

The assistant should run the game included with my thesis using the sample resources files. The level instructions for that game are included in the wiki/ subfolder of the thesis archive you downloaded.

Troubleshooting

All known common problems are in the troubleshooting doc on the repository wiki.

Possible improvements

Since the assistant needed to be ready and functional by a deadline, there are some features that would have been nice to have, but weren't necessary at the moment. Those features include:

• A "manual mode" where the user gets to decide next level

This would be similar to the original assistant prototype, where the user had the teacher tell them what branch they were choosing. It would also simplify some testing scenarios.

• A "dry run mode" for testing

Sometimes, it'd be nice to turn off the assistant's underlying calls to Vagrant, to speed up testing of features, config files, etc.

• The ability to turn off certain modules

In a game without flags, or where flag checking isn't important, the flag checker could be turned off, for example.

• Better support for branching levels -- arbitrary names, shorter and longer playthroughs

The assistant as it is right now expects the levels to have a naming convention, and it also expects that for every possible playthrough, there will always be the same number of levels. Removing these constraints would make it possible to support games of variable length, or with entire "replacement" branches, etc.