design-pattern-advisor
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a 'Base Framework' comprising multiple security policies (BF-SECURITY-1, BF-UNTRUSTED-1) that enforce strict evidence handling and prevent the agent from treating repository files as authoritative instructions.
- [SAFE]: Explicit instructions in the Untrusted-content policy (BF-UNTRUSTED-1) require the agent to ignore any prompt-injection directives found in data or artifacts.
- [SAFE]: The Command-execution policy (BF-COMMAND-1) restricts the agent from executing unknown scripts, fixture code, or passing secrets to untrusted environments.
- [SAFE]: The included validation script (scripts/validate_scenarios.py) is a benign static check for ensuring the skill contains necessary sections and scenarios; it does not execute untrusted code or access the network.
- [SAFE]: No obfuscation, hardcoded credentials, or unauthorized remote code execution patterns were identified in the provided files.
Audit Metadata