Rarely useful for incidents in my experience, mostly useful for onboarding those new to a codebase or part of a system. It’s a mechanism to preserve institutional knowledge that would otherwise evaporate due to not being represented as code; formalized documentation.
The benefit of this information existing in markdown files is this can also be used with LLMs and RAG if getting a natural language interface to the knowledge might be relevant to your enterprise.
That’s helpful context, thanks.
In practice, when incidents happen, where do people actually go to reconstruct the “why”? Is it mostly Slack archaeology and asking senior engineers, or is there something else that works better?
The benefit of this information existing in markdown files is this can also be used with LLMs and RAG if getting a natural language interface to the knowledge might be relevant to your enterprise.