skills/imraywang/rayskills/ray-cover/Gen Agent Trust Hub

ray-cover

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local Python scripts (brief_check.py and compose_cover.py) to validate design tasks and perform image composition. These scripts use standard libraries (json, pathlib, Pillow) and operate within the user-defined workspace.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted article text to extract visual metaphors and titles for image generation.
  • Ingestion points: Article content is read in SKILL.md (Step 1).
  • Boundary markers: The skill relies on analytical extraction steps rather than explicit delimiters, but the extracted content is validated against a strict schema.
  • Capability inventory: The skill performs file writes (images and JSON manifests) and runs local Python scripts.
  • Sanitization: Content is validated for length and forbidden keywords before being passed to the drawing script.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 02:46 PM
Security Audit — agent-trust-hub — ray-cover