aidlc-design
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: Benign instructional structure. The skill defines a strict multi-step process for architectural design, incorporating user checkpoints and validation rules. It does not contain any patterns typical of jailbreaks or safety overrides.
- [DATA_EXFILTRATION]: Safe implementation. The skill interacts with the local file system to manage design artifacts and uses external registry lookups (npm, PyPI, etc.) only for dependency version pinning. This behavior aligns with standard development tool functionality and follows the [TRUST-SCOPE-RULE] for well-known services.
- [COMMAND_EXECUTION]: No dangerous command execution patterns were found. The skill describes setup and build commands as part of the implementation design output for engineers but does not attempt to invoke these commands itself.
- [SAFE]: The skill implements security best practices, such as instructing the agent to use safe placeholders for environment variables and requiring explicit user approval before proceeding between design phases.
Audit Metadata