Introduction

@haskou/metrics counts attempted calls and failures, measures elapsed and CPU time, samples process memory, and emits structured lifecycle logs. It does not select a metrics backend, depend on a framework, or inspect method arguments and return values.

Principles

• metrics, logging, clocks, and resource sampling behind explicit ports
• function and decorator APIs over the MetricsInstrumenter class
• no DDD or framework assumptions
• Value Objects for validated numeric values, operation names, and stack traces
• cohesive measurement, resource-snapshot, and log objects at port boundaries
• no automatic argument or return-value logging
• deterministic test doubles included under @haskou/metrics/testing

The default runtime stores telemetry in two bounded in-memory buffers. You can try the decorator before choosing a backend or logger.

Adapter and clock failures are contained. Instrumentation cannot turn a successful application operation into a failed one.