Usage Analytics & Monitoring

The dashboard provides observability and usage transparency. Metrics come from infrastructure-level telemetry. They do not depend on client-side estimation.

What you can observe

The dashboard surfaces two operational views:

• Request and usage metrics over time.

• Event delivery and outcome status for integrations.

Request & usage metrics

The dashboard surfaces aggregated request metrics across time windows. This is designed for trends and incident triage. It is not a replacement for raw request logs.

Common metrics include:

• Request volume.

• Error rate.

• Endpoint-level usage distribution.

Common breakdowns include:

• By API domain and endpoint family.

• By response class (2xx, 4xx, 5xx).

• By API key or key label.

How to use the metrics

Use metrics to:

• Detect abnormal usage patterns.

• Optimize request behavior.

• Plan upgrades proactively.

Spikes in 4xx errors often indicate auth, scope, or quota issues. Spikes in 5xx errors often indicate upstream instability or timeouts.

Metric semantics (deep dive)

• Metrics are computed server-side at request entry and completion.

• Aggregation uses fixed time buckets for consistent rollups.

• Some data may arrive late due to batching.

• Sensitive headers and secrets are excluded from telemetry payloads.

Event & alert visibility

For users leveraging webhooks, WebSockets, or x402 payments, the dashboard provides visibility into operational outcomes:

• Event delivery status.

• Failed webhook attempts.

• Payment-related request outcomes.

This enables operational debugging without direct log access.

Delivery semantics (deep dive)

Common attributes tracked per delivery attempt:

• Timestamp and attempt number.

• Outcome class (delivered, failed, retried).

• Response status class for webhook destinations.

• A non-sensitive correlation identifier for support and triage.

Retry behavior is typically exponential backoff with a capped attempt count.

Debugging a failing webhook

1

Confirm delivery state

Check whether the event is pending, delivered, or failed.

2

Inspect attempt outcomes

Look for HTTP status codes, retry attempts, and last error class.

3

Validate the receiving service

Confirm your endpoint can handle timeouts, retries, and idempotency.

Webhook bodies and secrets should be treated as sensitive. Avoid copying payloads into non-secure systems.