Service Catalog

During an incident, "who owns this" is a question that should never require asking around. A service catalog exists to make ownership, on-call routing, and dependencies a lookup instead of a Slack archaeology exercise. The catalog fails the moment it drifts from reality — a stale entry is worse than no entry, because it sends the incident responder to the wrong team with false confidence.

Treat the catalog as the source of truth by construction, not by discipline: an entry that isn't enforced by tooling will go stale, and a stale catalog is a liability wearing the costume of documentation.

1. Make registration a side effect of scaffolding, not a separate step

If registering a new service in the catalog is a manual form someone fills out after the fact, it gets skipped under deadline pressure and the catalog silently loses coverage. Wire catalog registration into the scaffolding path itself — creating a service from a template in golden-paths should register it automatically, with owner and metadata pre-populated from the template's prompts.