Demonstrates the framework by mapping CSA AI Controls Matrix (AICM v1.0.2) controls to FAIR-CAM categories. Runs all 5 dimensions against 20 real AICM controls. The mappings in this example are illustrative — they show how to wire the framework, not how AICM controls should be mapped to FAIR-CAM. Replace the taxonomy, expert mappings, and transitive crosswalk with your own for your use case.
This is the worked example described in Validating LLM-Generated Control Mappings Beyond Aggregate Accuracy (Cloud Security Alliance).
python examples/aicm_to_faircam.pyThe example loads examples/faircam_taxonomy.json, which is not included in the repository. Create this file with the following schema:
{
"DOMAIN_CODE": {
"name": "Domain Name",
"subdomains": {
"Subdomain Name": {
"functions": ["Function A", "Function B"]
}
}
}
}Each top-level key is a domain code. Each domain has a name, subdomains, and each subdomain has a list of functions. The example expects three domains (LEC, VMC, DSC) matching the FAIR-CAM taxonomy.