architectural-design
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a framework for architectural reasoning. It explicitly instructs the agent to respect safety, data-loss, and incorrect-outcome risks, and emphasizes modeling trust boundaries when dealing with credentials, sensitive state, or external tools.
- [NO_CODE]: The skill consists exclusively of instructional Markdown files and a JSON evaluation file. There are no shell scripts, Python files, or executable components that could perform unauthorized actions on the host system.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process architectural requirements and existing documentation (ingestion points in SKILL.md Step 1). It mitigates potential risks by providing a structured methodology that requires the identification of trust boundaries, enforcement mechanisms, and explicit evidence for security claims (SKILL.md Step 9). The instructions emphasize technical realization over product semantics, reducing the surface for behavioral manipulation through injected data.
Audit Metadata