What do you still gather manually before your coding agent can debug a production issue?

I've been building DebugBundle around a simple idea: give a coding agent the evidence it needs in one structured bundle, without making it navigate a complex monitoring setup.

For me, the interesting part is keeping that context simple and useful, rather than just giving the agent more data.

Thinking about the last production issue you investigated with an agent, what did you still have to find yourself? The relevant logs, which version was deployed, a config change, or something else?

I'd genuinely like to hear where you got stuck 🙂
I'm in "collecting as much feedback as possible" mode 😄