API Reference

MetricFlow provides three core metric types and a set of global functions for initialization and lifecycle management.

Initialization

mf = metricflow.init(
    exporter="stdout",     # Exporter backend
    interval=10,           # Export interval in seconds
    prefix="myapp",        # Optional metric name prefix
)

Parameter	Type	Default	Description
`exporter`	`str`	`"stdout"`	Exporter backend name
`interval`	`int`	`10`	Export interval in seconds
`prefix`	`str`	`""`	Prefix for all metric names
`labels`	`dict`	`{}`	Default labels for all metrics

Counter

A monotonically increasing value. Use for counting events like requests, errors, or completions.

counter = mf.counter(
    "request_count",
    description="Total HTTP requests",
    labels=["method", "status"]
)

Method	Description
`counter.inc(**labels)`	Increment by 1
`counter.add(value, **labels)`	Increment by given value
`counter.get(**labels)`	Get current value

counter.inc(method="GET", status="200")
counter.add(5, method="POST", status="201")

Gauge

A value that can go up or down. Use for measurements like temperature, memory usage, or active connections.

gauge = mf.gauge(
    "active_connections",
    description="Currently active connections",
    labels=["pool"]
)

Method	Description
`gauge.set(value, **labels)`	Set to specific value
`gauge.inc(**labels)`	Increment by 1
`gauge.dec(**labels)`	Decrement by 1
`gauge.get(**labels)`	Get current value

Histogram

Samples observations and counts them in configurable buckets. Use for measuring distributions like request latencies or response sizes.

hist = mf.histogram(
    "request_duration_seconds",
    description="Request latency distribution",
    labels=["endpoint"],
    buckets=[0.01, 0.05, 0.1, 0.5, 1.0, 5.0]
)

Method	Description
`hist.observe(value, **labels)`	Record an observation
`hist.get_count(**labels)`	Get total observation count
`hist.get_sum(**labels)`	Get sum of all observations

Global Functions

Function	Description
`metricflow.init(**kwargs)`	Initialize MetricFlow instance
`mf.flush()`	Force export all pending metrics
`mf.shutdown()`	Gracefully shutdown, flush remaining metrics
`mf.get_metrics()`	Return all registered metrics as dict