I have been working with using playbooks to assist the analyst by automating command line explanations. As an example we have some alerts that are built out to look to certain command line behaviors. The idea is to take that command line and feed it to Gemini to return an explanation and a verdict.
The playbook and the results look really good. The issue that I see is we are having about a 20% failure rate on the API endpoint returning a 503 error. We are only doing this in Dev, and the volume is very low.
While I concept looks valid, it seems the infrastructure will not support our initiative. Has anyone else worked with this or seen this behavior?
Also, there should is no Gemini Tag that I can add to this.
