Skip to main content
Use the destination API to save encrypted configuration for Splunk, Datadog, Elastic or an HTTPS webhook. First list destinations: delivery_available must be true before you enable delivery. A deployment without the worker returns 409 DELIVERY_UNAVAILABLE when you request enabled: true.

What is delivered

An enabled destination receives new audit events whose workspace and environment were verified when recorded. A production destination never receives sandbox events. Historical events and events without reliable environment attribution are not assigned a guessed environment or streamed. This does not promise that every legacy producer records every platform action; use recorded audit/export evidence to identify coverage gaps. Configuration is not proof of ingestion. status: enabled means the destination is enabled. The list response includes environment-wide delivery_counts for pending, delivered, failed and cancelled records. A delivered record has received the provider’s HTTP acknowledgement; downstream indexing, alerts and retention must also be checked in your SIEM.

Configure the receiver

Use a public HTTPS hostname on port 443, without credentials in the URL, query parameters or fragments. Private addresses, redirects and non-HTTPS endpoints are rejected. Each workspace environment supports up to 20 destinations. Provider contracts: Splunk HEC, Datadog logs, Elasticsearch indexing. Protocol support does not establish that your provider account has accepted an event; test the configured account. Supply secret for webhooks or token for the other providers, never both. Credentials are encrypted at rest and never returned. Changing an endpoint requires providing its credential again. Use a dedicated credential with ingestion-only privileges where supported.

Verify webhook signatures

The JSON envelope includes id, tenant_id, environment_id, occurred_at and event. The X-Truthlocks-Event-ID header repeats the event ID. Verify the exact HTTP body bytes before parsing or processing them:
Accept only the expected workspace/environment and record event IDs durably to reject replays. Return a 2xx response only after accepting the event into durable storage. Use constant-time signature comparison; do not log the shared secret or authentication headers.

Retries and configuration changes

Delivery is at least once. The queue is committed in the same database transaction as the attributed audit event, so an audit rollback cannot leave a delivery behind. A worker crash after receiver acceptance can cause a duplicate; deduplicate using the stable event ID. Transport failures and non-2xx responses retry with exponential backoff starting at one minute, capped at one hour, up to 20 attempts. Exhausted records remain failed; there is no automatic unlimited retry or promised completion time. Disabling or changing a destination cancels its pending deliveries to avoid forwarding an old backlog to a changed receiver. Deletion removes configuration, the saved credential and its delivery queue records. Preserve required delivery evidence before deletion. There is no automatic historical replay.

Snapshot exports

Use audit exports for recorded audit snapshots and risk-event exports for supported risk-signal snapshots. Persist processing position and deduplicate event identities when importing snapshots. Export jobs, SIEM delivery and downstream alerting are separate operations; completing one does not prove the others succeeded.