Skills
skills/flight505/storybook-assistant/visual-design/Gen Agent Trust Hub

visual-design

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by incorporating project data into AI prompts. \n
  • Ingestion points: User-supplied descriptions and project metadata (framework, design system, colors, fonts) referenced in SKILL.md and scripts/generate_mockup.py. \n
  • Boundary markers: Absent. Contextual data is directly interpolated into prompt strings. \n
  • Capability inventory: The skill uses Bash to run scripts and Write to save generated images to the local disk. \n
  • Sanitization: External project context is not sanitized or escaped before being sent to the AI model. \n- [EXTERNAL_DOWNLOADS]: The Python script scripts/generate_mockup.py makes network requests to the well-known OpenRouter API at https://openrouter.ai. This is necessary for its core functionality of generating AI images. \n- [COMMAND_EXECUTION]: The skill relies on executing local Python scripts via the Bash tool to handle AI model interaction and image processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 05:50 AM