css-facsimile

Pass

Audited by Gen Agent Trust Hub on Sep 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external document transcripts and scan data, which creates a surface for indirect prompt injection.
  • Ingestion points: SKILL.md identifies inputs as "per-folio verbatim transcripts" and "original scans."
  • Boundary markers: The document does not specify technical delimiters for these inputs, relying on workflow staging.
  • Capability inventory: The skill orchestrates the generation of PDF/PNG files and utilizes local scripts for processing.
  • Sanitization: No automated sanitization is described; the process relies on an "Orchestrator" role to visually verify outputs against original scans.
  • [COMMAND_EXECUTION]: The instructions reference the use of project-specific scripts and standard CLI utilities.
  • Evidence: The text mentions v11n3-run/measure.py for px-to-pt arithmetic and pdfunite for merging final documents. These are standard tools for the stated typesetting purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 30, 2026, 12:07 PM