event-driven-architecture

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill describes patterns for ingesting data from external message brokers (Kafka and SQS) which creates a surface for indirect instructions.
  • Ingestion points: The Kafka consumer example in SKILL.md uses JSON.parse on external message values.
  • Boundary markers: The example does not implement delimiters or explicit "ignore instructions" logic for the event payload.
  • Capability inventory: The skill performs database writes via Prisma and executes domain-specific business logic handlers based on ingested event types.
  • Sanitization: No explicit validation, schema enforcement, or sanitization of the event payload is demonstrated in the processing examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 08:36 PM
Security Audit — agent-trust-hub — event-driven-architecture