compliance-management

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides comprehensive logic for tracking regulations, certifications, and trade compliance without using any sensitive system calls, network communication, or persistent file operations.\n- [PROMPT_INJECTION]: The skill's functions process data that could be influenced by external actors, creating an indirect prompt injection surface.\n
  • Ingestion points: The add_product, add_supplier, and classify_product methods in SKILL.md accept arbitrary string inputs for product descriptions, supplier names, and material compositions.\n
  • Boundary markers: There are no boundary markers or explicit instructions provided to the agent to ignore potentially malicious instructions embedded within these data fields.\n
  • Capability inventory: The skill contains no dangerous capabilities; it lacks any code to perform network requests, spawn subprocesses, or write to the local file system.\n
  • Sanitization: The implementation does not include any validation, escaping, or sanitization logic for the string inputs processed by the compliance and trade management classes.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 08:19 PM