skills/canine89/hwpxskill/hwpx/Gen Agent Trust Hub

hwpx

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill uses established Python libraries and standard file operations to perform document processing tasks locally. No malicious obfuscation, persistence, or data exfiltration patterns were identified.
  • [PROMPT_INJECTION]: An indirect prompt injection surface exists due to the processing of untrusted document content.
  • Ingestion points: Content is read from external HWPX files in scripts/analyze_template.py and scripts/text_extract.py.
  • Boundary markers: Content is extracted and processed based on the OWPML schema without additional isolation markers.
  • Capability inventory: The skill enables the agent to execute internal scripts for file modification and ZIP management.
  • Sanitization: No specific filtering or escaping is applied to extracted document text, as the primary purpose is document reconstruction.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 09:00 AM