ecommerce-image-workflow

Pass

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes a bash script in SKILL.md to automate calls to the $OD_BIN media generation tool and uses python3 -c for JSON output parsing. These commands are integral to the image generation workflow and utilize platform-provided environment variables.- [PROMPT_INJECTION]: The skill processes external image data, which introduces a potential surface for indirect prompt injection.
  • Ingestion points: User-provided reference images (SKILL.md, Step 0).
  • Boundary markers: The skill uses a mandatory fidelity lock (Step 3) that instructs the agent to preserve shape, color, and material, acting as a behavioral constraint.
  • Capability inventory: The skill can execute shell commands and Python scripts via Step 4 to manage media assets.
  • Sanitization: No explicit validation or filtering is performed on the visual contents of the uploaded reference images.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 7, 2026, 02:57 AM
Security Audit — agent-trust-hub — ecommerce-image-workflow