recoup-internal-consulting-article-illustrator

Pass

Audited by Gen Agent Trust Hub on Oct 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill workflow requires the agent to execute a local Node.js script (brand/materialize.mjs) to prepare the branding environment. The script performs file operations including directory creation and asset copying within the workspace.
  • [DYNAMIC_EXECUTION]: The materialize.mjs script dynamically creates a markdown configuration file (frame.md) and an HTML template (caption-skin.html) at runtime. This process involves reading local CSS and font files and embedding the font data into the generated HTML.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes article text to generate visual concepts. Since it lacks explicit delimiters or instructions to ignore commands within the source text, it is theoretically susceptible to instructions hidden in the content influencing the image generation prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 10, 2026, 02:15 PM