shard
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides instructional architectural templates and SQL scripts. It contains no executable scripts, binaries, or automated installation processes.\n- [SAFE]: All external URLs point to well-known technical documentation and official blogs from service providers like PostgreSQL, Microsoft, and Supabase, providing legitimate context for multi-tenant design.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied requirements and schemas, which constitutes a surface for untrusted data ingestion.\n
- Ingestion points: User-provided SaaS requirements and database schemas (SKILL.md).\n
- Boundary markers: The skill mandates 'fail-closed' security policies and the use of transaction-scoped context markers to prevent cross-tenant leakage.\n
- Capability inventory: The skill is limited to generating documentation and SQL code; it lacks tools for network operations, filesystem access, or command execution.\n
- Sanitization: The design workflow incorporates mandatory data leakage checklists and architectural reviews to ensure security constraints are applied to all generated designs.
Audit Metadata