Using LLMs against Fortigate configurations (claude.ai, gemini, Le Chat, DeepSeek, others)
Anybody started dabbling with LLMs (your flavour) and Fortigate configs?
I was wondering if they could be useful in the following contexts:
- running a check against a baseline configuration (settings sync between multiple FGTs)
- looking for and report sub-optimal or unsafe settings as per current best practices
- looking for signs of past compromises (excess local admins, other undesired modifications as per published past PSIRT / CVE indicators)
Are we near a useful level of maturity LLM-side, or still too many hallucinations and lack of consistency?
[EDIT: lots of interesting suggestions incoming... looks like an hot topic!]