data-model-creation
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill explicitly instructs the agent to avoid fetching remote skill definitions or protocol markdown from the internet, requiring that sibling skills be accessed only via local relative paths. This proactive constraint prevents remote code execution and data exfiltration through unauthorized external connections.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided business descriptions to generate Mermaid diagrams and corresponding data models. While this involves an ingestion surface for untrusted data, it represents the primary intended function of the skill and does not provide an escalation path for malicious commands. 1. Ingestion points: User requests describing business entities, ownership, and cardinality (SKILL.md). 2. Boundary markers: No explicit markers or ignore-instructions are defined for the modeling workflow. 3. Capability inventory: Use of manageDataModel and modifyDataModel tools to interact with database schema structures (SKILL.md). 4. Sanitization: The instructions provide naming rules (e.g., PascalCase) but do not detail sanitization of user-provided business descriptions before model generation.
Audit Metadata