safe-ai-dlc
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown and does not include any scripts, commands, or automated execution patterns. It provides a structured workflow for manual or agent-assisted project tracking.
- [EXTERNAL_DOWNLOADS]: The skill references an external methodological guide from the official AWS blog. This targets a well-known technology service and is documented for background information rather than automated execution.
- [DATA_EXFILTRATION]: No patterns of data exfiltration were found. The instructions explicitly define a Human-Gate Checklist that requires human approval for any actions involving credentials, secrets, or security policies, preventing autonomous unauthorized access.
- [SAFE]: The skill has an ingestion surface for processing audits and epics. 1. Ingestion points: Processing audit findings into tracker structures as described in SKILL.md. 2. Boundary markers: Explicit requirements for human validation of business and security context. 3. Capability inventory: Tracker operations via external SOPs. 4. Sanitization: All high-risk decisions (credentials, security policies) are routed to a human for validation.
Audit Metadata