owasp-security

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFE
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill contains examples of unsafe functions such as eval(), exec(), and os.system(). These are provided solely for educational purposes within markdown code blocks to demonstrate vulnerabilities and their safe alternatives. The skill itself does not utilize these functions for its own logic.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines guidelines for processing external code and data, which is an inherent attack surface for agentic AI. However, it explicitly provides remediation strategies, such as input validation, sanitization, and the use of safe APIs, to mitigate these risks and does not introduce injection vulnerabilities within its own instructions.
  • [CREDENTIALS_UNSAFE]: No hardcoded credentials or secrets were found. The skill includes explicit instructions for developers to store secrets in secure environments or vaults rather than in the code itself.
  • [PROMPT_INJECTION]: No malicious prompt injection patterns or attempts to override agent behavior were detected. The use of emphasis markers like 'Important' and 'Critical' is consistent with standard technical and security documentation.
  • [DATA_EXFILTRATION]: The skill does not perform any network operations, access sensitive system files, or exfiltrate data. It is an instructional reference guide focused on providing security best practices for application development.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:07 PM