Developer DocsstableUpdated 2026-07-06

Logs & Traces

Search Gateway requests, direct ingestion attempts, webhook deliveries and developer signals from one operational view.

Logs & Traces is the developer investigation surface for Kadryn traffic.

Use it to answer:

  • did Kadryn receive the request?
  • did Gateway forward it?
  • did direct ingest accept or reject it?
  • did a webhook deliver successfully?
  • what provider, model, project and environment were involved?
  • what trace ID or request group ID connects this event to my app logs?

What appears here

Depending on your workspace, Logs & Traces can show:

  • Gateway requests;
  • direct ingestion attempts;
  • webhook deliveries;
  • policy decisions;
  • fallback usage;
  • errors;
  • synthetic tests;
  • expensive requests;
  • request groups;
  • trace IDs.

What to search by

Useful search fields include:

  • Kadryn request ID;
  • trace ID;
  • request group ID;
  • provider request ID;
  • project;
  • feature;
  • environment;
  • provider;
  • model;
  • status;
  • error code;
  • webhook endpoint.

Common filters

Use filters to narrow the investigation:

FilterUse it when
GatewayYou are debugging runtime provider traffic.
Direct ingestYou are debugging usage import.
WebhooksYou are debugging deliveries or redeliveries.
Errors onlyYou need failures first.
Expensive onlyYou are investigating high-cost calls.
Fallback usedYou suspect provider fallback or routing changes.
Synthetic onlyYou are validating tests rather than production traffic.

Gateway investigation

When a Gateway request fails:

  1. search by trace ID or request group ID;
  2. inspect status and error code;
  3. check provider route status;
  4. check model and provider;
  5. review policy decision if blocked;
  6. preserve Kadryn request ID in support notes.

Direct ingest investigation

When an ingestion event is missing:

  1. search by idempotency key, provider request ID or trace ID;
  2. check accepted vs rejected attempts;
  3. inspect validation errors;
  4. verify timestamp and environment;
  5. check whether the event appears in Costs after rollup.

Webhook investigation

When a webhook delivery fails:

  1. search by endpoint or delivery ID;
  2. inspect response status;
  3. check retry state;
  4. verify signature handling;
  5. redeliver only after fixing the receiver.

Synthetic traffic

Synthetic traffic is useful for validation.

Use it to verify:

  • API key setup;
  • Gateway readiness;
  • provider route readiness;
  • webhooks endpoint readiness;
  • log visibility;
  • trace search.

Do not treat synthetic traffic as proof of real production traffic.

Logging guidance

Keep these IDs in your application logs:

  • trace ID;
  • request group ID;
  • Kadryn request ID;
  • provider request ID;
  • idempotency key hash or prefix;
  • project;
  • environment.

Do not log secrets or full Authorization headers.