solution-design
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a set of 'Hard Rules' that strictly define the agent's behavior, including prohibitions against inventing business rules, assuming architectures, or claiming unperformed validations. These constraints ensure the agent remains a passive design assistant.
- [INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection as it is instructed to inspect existing repositories (structure, documentation, and code).
- Ingestion points: Local repository files including structure, documentation, configuration, models, migrations, and tests (SKILL.md, Hard Rule 2).
- Boundary markers: Explicit instructions to separate verified facts, accepted decisions, proposals, and assumptions (SKILL.md, Hard Rule 3).
- Capability inventory: File system reading (for inspection) and creation of Markdown documentation files (Execution Steps).
- Sanitization: The skill instructions emphasize qualitative analysis and documentation generation rather than automated code processing or execution, which naturally limits the impact of data-driven injections.
Audit Metadata