Post #4390574

2026-08-05 05:41 UTC

@tbortels@infosec.exchange @mhoye@cosocial.ca It's unclear to me exactly what kind of useful patterns the LLM is "inferring" for you, and what useful output it is producing as a result of this "inferring". You mention python and log records being different based on event types. Do you mean the LLM is outputting python code that branches on say a `"type"` or `"kind"` field in the JSON?

Replies (1)

@tbortels@infosec.exchange 2026-08-05 05:57

@mhoye@cosocial.ca @Ryan1729@mastodon.social It can. But I don't need to get that specific. "Write a python script to run daily that goes to cloudwatch for the last day in account xyz in us-west-2, and give me a summary of failure modes for the top ten mentioned lambdas" mostly just works. And I didn't need to mess with pagination on cloudwatch or escaping or decoding or working out the jsonpath. And that prompt will work out-of-the-box 9 out of 10 times. And then I can say "good, extend that for all active regions" - and it just works. Minutes instead of an hour of screwing around. Now I can spend my time on interesting, non-routine tasks. The reality - jq users will know this - is that jq syntax is difficult and quirky and simple things are doable but complex things are difficult. Beyond a certain complexity - just go do python. It's easier to code and easier to read. But neither of them will understand from context when I say "include any tags that are relevant".

Open ##4390573