quality-documentation-manager

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill comprises documentation guides and local Python scripts designed to facilitate ISO 13485 and 21 CFR Part 11 compliance.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes scripts that process document metadata in JSON format, which represents a potential ingestion surface.\n
  • Ingestion points: Metadata files processed via scripts/document_validator.py and scripts/document_version_control.py.\n
  • Boundary markers: The tools interpret structured JSON data rather than natural language instructions contained within the documents.\n
  • Capability inventory: Capabilities are restricted to local file operations (read, write, rename) and SHA256 hashing. No network access or system command execution functionality is included.\n
  • Sanitization: The scripts utilize standard library parsing and regular expression validation for document numbering formats.\n- [DATA_EXFILTRATION]: No network communication or access to sensitive system paths was detected.\n
  • Evidence: The provided scripts do not import network-capable libraries such as requests, urllib, or socket. All file operations are confined to the local project directory and specified storage paths.\n- [COMMAND_EXECUTION]: Python scripts perform standard file system management tasks for version control.\n
  • Evidence: The tools use pathlib, json, and os.rename for document archiving. They do not employ dangerous execution functions like eval(), exec(), or subprocess modules to run arbitrary shell commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 02:33 PM
Security Audit — agent-trust-hub — quality-documentation-manager