privacy-us

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or security vulnerabilities were detected in the skill instructions or structure.
  • [DATA_EXFILTRATION]: The skill collects business-identifying information (entity name, address, email) through a guided interview to populate privacy policy templates. This data is used solely for the purpose of document generation and is not exfiltrated to external domains.
  • [COMMAND_EXECUTION]: No unauthorized or dangerous command execution was found. The skill generates static MDX and TSX files in the project's source directory and does not involve arbitrary shell commands or privilege escalation.
  • [PROMPT_INJECTION]: The skill processes user-provided input during the interview stage to populate legal templates. 1. Ingestion points: User responses to interview questions Q9US-1 through Q9US-10 (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Writing generated files to the local filesystem in the src/ directory (SKILL.md). 4. Sanitization: Absent. This represents a standard document generation surface with no evidence of malicious intent or safety bypass instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 05:04 AM
Security Audit — agent-trust-hub — privacy-us