Observability & Instrumentation
Instruments code so production behavior is visible and diagnosable — telemetry written alongside the feature, the same way tests are, so the first user-reported bug is a query and not archaeology. Its distinctive discipline is question-first: before adding any signal, write the 2-4 questions an on-call engineer will ask, then map each signal to one. It picks the right signal per question — metrics say that something is wrong, traces say where, logs say why.
Concretely it prescribes structured logging with stable event names and mandatory correlation IDs (never secrets or PII), RED metrics on every endpoint and dependency with bounded label sets (cardinality is the failure mode), OpenTelemetry distributed tracing, and symptom-based alerting on what users feel rather than on causes like CPU. Observability is a new domain for this wiki; it feeds the launch-day monitoring that addy-shipping depends on.
See Also
- addy-shipping — the launch skill whose monitoring this instrumentation feeds.
- stage-release — the canonical stage this implements.