We expanded the observability surface around RAG infrastructure. The codebase includes health and metrics modules so runtime state can be inspected instead of inferred.
This does not claim production uptime or performance. It establishes the plumbing required to measure and debug those properties.


