SAFT-GT Pipeline

The SafeSec Attack-Fault Tree Generation Toolchain (SAFT-GT) was created in the SafeSec Project. It generates and analyzes Attack-Fault Trees for cyber-physical systems based on their data flow and the hardware and software components used. Vulnerability databases are crawled to identify possible attacks.

If you use the SAFT-GT pipeline, models, or any part of this repository in your research, please kindly cite our paper:

@article{pekaric2025saftgt,
  title   = {Bridging Safety and Security in Complex Systems: A Model-Based Approach with SAFT-GT Toolchain},
  author  = {Pekaric, Irdin and Groner, Raffaela and Raschke, Alexander and Witte, Thomas and Adigun, Jubril Gbolahan and Felderer, Michael and Tichy, Matthias},
  journal = {Journal of Systems and Software},
  year    = {2026}
}

Contents

The content of this supplement package is structured as follows:

• /models
  contains models used or created by SAFT-GT for our example
• /saft-msgs
  the definition of the ROS messages used to trigger the different phases and to send the calculated MTTF
• /saft_pipeline
  the pipeline integrated in a ROS node
• /scripts
  scripts used in the Docker container for performance measurement and testing.
• /tools
  the tools needed for the different stages of the pipeline:
  • /tools/data2deploy.jar extracts an initial deployment model from a given dataflow model
  • /tools/dependency_explorer extracts the dependencies of the components in a given deployment on a specific system
  • /tools/attackgraphgenerator
    tries to find vulnerabilities for given components
  • /tools/AFTGenerator.jar
    generates an AFT from FTs, ATs, Dataflow and Deployment models
  • /tools/SAFTProject.jar
    contains dependencies needed by AFTGenerator.jar
  • /tools/aft2dft.jar
    converts an AFT to a DFT

Besides the pipeline itself, we have added the following additional files

• /grammars
  contains the Xtext grammars of all introduced domain-specific languages (DSLs)
• /workshop_presentations
  contains the presentations given at the workshop with experts to evaluate our approach, transcript of discussions, linked survey questions and consent forms.
• /saft-gt-evaluation_rawdata.xlsx contains the raw results of the performance measurement

In the root folder, a Dockerfile can be used to create a ROS container that runs the pipeline in a ROS environment. This should be started along with the Storm model checker most easily with docker compose and the docker-compose.yml file.

Remarks

• It is not possible to run this package from scratch! You'll need a running ROS system that can be used to extract the dataflow and can be accessed by the dependency_explorer to extract the dependencies of the ROS nodes on this particular system. Due to Hardware and OS depedencies, it wouldn't make sense to provide this system.
• The atttackgraphgenerator can be executed stand-alone by first starting a DB-container with the /tools/attackgraphgenerator/docker-compose.yml-file and then fill the database by calling ./init.py.
  It might be necessary to add an API key to a file api.json with the content {"key" : "<API-Key goes here>"}
  After this initization, the attackgraphgenerator can be used by calling ./generate.py with the parameters -p or -s followed by a string that is searched only in the CPEs resp. in the descriptions. A kind of "batch-processing" is possible by using the parameters -fp or -fs followed by a filename containing a set of strings to be searched, one per line.

Contact

If you want further information about this project, its results or the tool pipeline, don't hesitate to contact us via e-mail alexander.raschke@uni-ulm.de