I keep running into the same pattern: incident happens, go to debug, realize there’s no logging for that code path. Add it after the fact, wait for it to happen again.

Thinking about building an open source tool that runs on PRs and automatically adds structured logging, metrics, and tracing. It would analyze your existing codebase to match your conventions.

The thing I find interesting: if a tool is adding all the instrumentation, it has a model of your entire system. From that you could generate service maps, dashboards, alerting — stuff that’s useful but nobody maintains.

Before I build this: is this actually a problem people have, or do mature teams already have this solved? Would love to hear how others handle this.