SIEM

The audit log is designed to be ingested into your Security Information and Event Management (SIEM) system.

Log Format

Audit entries are newline-delimited JSON (NDJSON), making them compatible with most log ingestion pipelines:

  • Splunk - Use the HTTP Event Collector or file monitoring
  • Elastic/ELK - Filebeat or Logstash can ingest directly
  • Datadog - Log collection agent supports JSON
  • AWS CloudWatch - Container stdout goes directly to CloudWatch Logs

Recommended Setup

For production deployments, we recommend:

  1. Output to stdout in your container:

    audit:
      path: stdout
    If your application logs also go to stdout, audit entries will be interleaved. To keep them separate, configure application logging to stderr via log.output: stderr so your log driver can split by stream. Alternatively, write the audit log to a file on a separate mounted volume and collect it independently.
  2. Let your orchestrator handle log collection - Kubernetes, ECS, and Docker Compose all have built-in log drivers that can forward to your SIEM.

  3. Configure your SIEM to parse JSON - The audit log is already structured, so minimal parsing is needed.

Key Fields for Alerting

Consider creating alerts for:

FieldConditionAlert
action"deny"Tool call was blocked
action_reason"fail_open"Validation timed out
duration_ms> 10000Slow validation
request_validation.ai.results[].errorexistsAI validation error