mal-simulator 2.5.0

A MAL compliant simulator.

MAL Simulator

Overview

The MAL Simulator is a tool designed for running simulations on systems modeled in the Meta Attack Language (MAL).

It allows you to:

• Simulate the adversarial and defensive actions of intelligent agents (attackers and defenders).
• Analyze system security and evaluate attack/defense strategies.
• Develop machine learning agents (e.g., using Reinforcement Learning) in a safe, dynamic environment.
• Incorporate stochastic modeling through Time-to-Compromise (TTC) distributions.

For in-depth documentation, refer to:

• The official Wiki.
• The official MAL infrastructure tutorials.

Contributing

• Use Conventional commits
• The CI pipeline runs pytest, mypy and ruff format && ruff check for linting and type checking. PRs will only be merged if pipeline succeeds.

Making a release

• Make a release PR (called release-x.y.z) with your changes. Update version number in pyproject.toml and malsim/__init__.py.
• Merge the PR after the pipeline suceeds
• Tag the latest commit with the new version number
• Push the tag

---

🛠️ How to Run a Simulation

1. Installation

pip install mal-simulator

To also get ML dependencies (pettingzoo, gymnasium):

pip install mal-simulator[ml]

For additional dev tools:

pip install mal-simulator[dev]

2. Running a Scenario via CLI

You can execute a simulation using a YAML scenario file, which defines the MAL model, the agents (attackers and defenders), their policies, and their goals.

The basic command structure is:

malsim <scenario_file> [options]

Example

Execute a simulation defined in my_scenario.yml:

malsim my_scenario.yml

Common Options:

Option	Description
-s, --seed SEED	Sets the random seed for deterministic simulations.
-t, --ttc-mode MODE	Sets the Time-to-Compromise mode (e.g., DISABLED, PER_STEP_SAMPLE).
-g, --send-to-gui	Sends simulation data to the malsim-gui for visualization (requires the GUI to be running).

3. Programmatic Usage

For advanced and customized simulations (e.g., integrating with a custom ML training loop), you can initialize and run the simulator directly in a Python script:

from malsim import MalSimulator, run_simulation, Scenario

# 1. Load the scenario configuration
SCENARIO_FILE = "my_scenario.yml"
scenario = Scenario.load_from_file(SCENARIO_FILE)

# 2. Create the simulator instance
sim = MalSimulator.from_scenario(scenario)

# 3. Run the simulation
agent_actions = run_simulation(sim, scenario.agents)

print("Simulation finished. Agent actions:", agent_actions)

You can also run your own simulation loop.

from malsim import MalSimulator, Scenario

scenario = Scenario.load_from_file("my_scenario.yml")
sim = MalSimulator.from_scenario(scenario)
agent_states = sim.reset()

while not sim.done():
    # ... calculate actions for each agent
    agent_states = sim.step(actions)

Read the Wiki for more ways to run the simulator.