aws-cloud-patterns
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [Prompt Injection] (SAFE): No instructions targeting agent behavior or safety overrides were detected. Content is purely technical documentation.
- [Data Exposure & Exfiltration] (SAFE): No hardcoded secrets or exfiltration patterns found. The skill explicitly lists hardcoding credentials as an anti-pattern.
- [Indirect Prompt Injection] (LOW): The code snippets demonstrate handling untrusted data from AWS events (S3, API Gateway). Evidence:
- Ingestion points: SKILL.md (Lambda and S3 handlers process event parameters).
- Boundary markers: Not applicable (documentation snippets).
- Capability inventory: Uses standard AWS SDK calls for DynamoDB and S3.
- Sanitization: Uses standard URL decoding for S3 keys.
- [External Downloads] (SAFE): No remote scripts are downloaded or executed. References to standard AWS SDKs and CDK libraries are for documentation purposes.
Audit Metadata