data-storage
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists exclusively of markdown documentation and interface configuration. No executable scripts, subprocess calls, or network operations are present. All referenced external links point to established standards bodies, official product documentation, or reputable research sources.
- [NO_CODE]: There are no code files or executables included with the skill. Logic is conveyed through natural language instructions for the agent to follow during analysis and implementation tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a rigorous framework for analyzing data storage changes. While it guides the agent in processing user-supplied schemas and data, it mandates the use of authoritative database constraints and explicit validation logic to maintain integrity.
- Ingestion points: User-provided database schemas and models (SKILL.md).
- Boundary markers: Present in the Domain Law section (SKILL.md).
- Capability inventory: Suggested schema and repository changes.
- Sanitization: Instructs the agent to map untrusted input to validation and authorization boundaries (Paper 018).
Audit Metadata