name: sentry-setup-metrics
description: Setup Sentry Metrics in any project. Use when asked to add custom metrics, track counters/gauges/distributions, or instrument application performance. Supports JavaScript and Python.

Setup Sentry Metrics

Configure Sentry's custom metrics for tracking counters, gauges, and distributions.

Invoke This Skill When

• User asks to "add Sentry metrics" or "track custom metrics"
• User wants counters, gauges, or distributions
• User asks about Sentry.metrics or sentry_sdk.metrics

Quick Reference

Note: Ruby does not have metrics support.

Metric Types

JavaScript Setup

Metrics are enabled by default in SDK 10.25.0+.

Counter

Sentry.metrics.count("api_call", 1, {
  attributes: { endpoint: "/api/users", status_code: 200 },
});

Gauge

Sentry.metrics.gauge("queue_depth", 42, {
  unit: "none",
  attributes: { queue: "jobs" },
});

Distribution

Sentry.metrics.distribution("response_time", 187.5, {
  unit: "millisecond",
  attributes: { endpoint: "/api/products" },
});

Filtering (optional)

Sentry.init({
  beforeSendMetric: (metric) => {
    if (metric.attributes?.sensitive) return null;
    return metric;
  },
});

Python Setup

Metrics are enabled by default in SDK 2.44.0+.

Counter

sentry_sdk.metrics.count("api_call", 1, attributes={"endpoint": "/api/users"})

Gauge

sentry_sdk.metrics.gauge("queue_depth", 42, attributes={"queue": "jobs"})

Distribution

sentry_sdk.metrics.distribution(
    "response_time", 187.5,
    unit="millisecond",
    attributes={"endpoint": "/api/products"}
)

Filtering (optional)

def before_send_metric(metric, hint):
    if metric.get("attributes", {}).get("sensitive"):
        return None
    return metric

sentry_sdk.init(dsn="YOUR_DSN", before_send_metric=before_send_metric)

Common Units

Timing Helper Pattern

JavaScript

async function withTiming(name, fn, attrs = {}) {
  const start = performance.now();
  try { return await fn(); }
  finally {
    Sentry.metrics.distribution(name, performance.now() - start, {
      unit: "millisecond", attributes: attrs,
    });
  }
}

Python

import time, sentry_sdk

def track_duration(name, **attrs):
    def decorator(fn):
        def wrapper(*args, **kwargs):
            start = time.time()
            try: return fn(*args, **kwargs)
            finally:
                sentry_sdk.metrics.distribution(
                    name, (time.time() - start) * 1000,
                    unit="millisecond", attributes=attrs
                )
        return wrapper
    return decorator

Best Practices

• Low cardinality: Avoid user IDs, request IDs in attributes
• Namespaced names: api.request.duration, not duration
• Flush on exit: Call Sentry.flush() before process exit

Troubleshooting