cloud-forensics
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill integrates with established cloud service SDKs including boto3, azure-mgmt, and google-cloud libraries. All demonstrated functionality aligns with the stated purpose of cloud forensics and security investigation. The handling of cloud credentials via environment variables and local profiles follows security best practices for administrative tooling.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted log data from external cloud environments, which represents a potential surface for indirect prompt injection.\n
- Ingestion points: External cloud logs retrieved via methods such as get_cloudtrail_events, get_activity_logs, and search_unified_audit_log as described in SKILL.md.\n
- Boundary markers: No specific delimiting or framing of untrusted log content is defined in the instruction file.\n
- Capability inventory: The skill includes methods for generating HTML/CSV reports and preserving cloud resource states (e.g., snapshots).\n
- Sanitization: No explicit mention of log content sanitization or validation logic is provided in the instruction set.
Audit Metadata