realistic-ugc-video

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a generation script located at ~/.claude/skills/nano-banana/scripts/generate.sh. This facilitates integration with the Nano Banana image generation skill for creating base images.
  • [INDIRECT_PROMPT_INJECTION]: The production workflow involves ingesting user-supplied scripts and descriptions which are then interpolated into prompts for external generation models like Kling AI.
  • Ingestion points: User-provided character descriptions and video scripts collected in Phase 1 of the workflow.
  • Boundary markers: None identified in the provided prompt templates; the system relies on structured instructions for the LLM to follow.
  • Capability inventory: The skill has access to the Bash tool for local script execution and the Task tool for spawning background subagents to handle video processing.
  • Sanitization: No specific sanitization or filtering of user input is documented before interpolation into generation prompts, which is standard for production-oriented creative tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 11:48 PM