Minimal working example to showcase the integration of a small ML classifier into the BICEP framework
In order to be able to start the project you will need to initialize it first. Do this by running:
git submodule update --init --recursive
This fetches the newest version of the submodule for the backend code and is necessary for the application to work seamlessly.
To build a local version of the image for testing purposes, simply run:
cd ./bicep-example
docker buildx build . -t <name>:<tag>
After building the docker image locally or pushing it to a registry, you will want to integrate it into a BICEP instance. To do this, either add it via the frontend in a running instance by navigating to IDS Tools and clicking + Add New Tool. Fill in the required parameters and select SINGLE_CONTAINER as Deployment Type and both for the analysis method. As this example does not need a ruleset, uncheck the "requires ruleset" box.
Please note that after a restart of the framework your added tool might be gone. Therefor, we recommend adding it via adapting the SQL script and rebuilding and deploying the database container.
In case you need advanced configuration options, you can also upload a configuration file to the BICEP framework. The file will be automatically injected to the running container. If you don't need one, you can specify during IDS setup any of the sample configuration files.