privacy-policy-stephane-boghossian

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill instructions include a 'prime directive' and 'hard rules' focused on maintaining accuracy and preventing the AI from fabricating legal citations. These internal guidelines serve functional safety purposes rather than attempting to subvert agent behavior.
  • [EXTERNAL_DOWNLOADS]: The skill mentions optional 'live grounding' using the legal-data-hunter and cite-guard tools. These appear to be Model Context Protocol (MCP) extensions or platform-specific tools for data verification rather than arbitrary downloads or untrusted code execution.
  • [DATA_EXFILTRATION]: No unauthorized network operations or data leakage patterns were identified. The skill generates text based on user-provided intake answers and verified reference files.
  • [COMMAND_EXECUTION]: The skill contains no shell commands, subprocess calls, or system-level operations. It focuses entirely on text assembly using markdown templates.
  • [DYNAMIC_EXECUTION]: The skill uses a template-based assembly process (assets/template-privacy-policy.md) to generate its output. This is a standard and safe method for document generation that does not involve executable code or unsafe deserialization.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 07:33 PM
Security Audit — agent-trust-hub — privacy-policy-stephane-boghossian