schema-evaluation
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No security issues detected. The skill is purely informational and provides a structured methodology for data modeling tasks without invoking external tools, scripts, or network resources.\n- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an ingestion surface for untrusted data (DDL, ERDs), though it possesses no high-risk capabilities to exploit.\n
- Ingestion points: Reads schema definitions, DDL, ERDs, and data dictionaries in SKILL.md.\n
- Boundary markers: Absent; inputs are handled as read-only analysis targets.\n
- Capability inventory: No subprocess calls, file writes, or network operations are defined in the skill.\n
- Sanitization: The skill explicitly states that inputs are treated as read-only and not used in command construction.
Audit Metadata