document-granular-decompose

Pass

Audited by Gen Agent Trust Hub on May 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes content from external documents (PDF, Word, etc.) and returns the full text to the agent, creating a surface for indirect prompt injection where a malicious document could contain instructions intended to override agent behavior.\n
  • Ingestion points: Untrusted data enters the agent context via the text extraction functionality in scripts/mineru_fulltext_extract.py which reads local files.\n
  • Boundary markers: The skill does not implement delimiters or 'ignore embedded instructions' warnings for the returned text.\n
  • Capability inventory: The skill script has file read, file write, and network POST capabilities (utilizing urllib.request).\n
  • Sanitization: No validation or filtering is applied to the extracted content before it is provided to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
May 13, 2026, 03:33 PM