storyboard
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- Command Execution: The skill involves running local utility scripts such as html2png.mjs and brand-lint.mjs, and CLI tools like hyperframes and video-gate within the /workspace/ environment. These commands are essential for generating and validating storyboard assets.
- Indirect Prompt Injection: The skill ingests data from project-specific files, which represents a potential surface for instruction injection if the files are attacker-controlled. (1) Ingestion points: storyboard.md, artifact-contract.json, and HTML templates. (2) Boundary markers: No explicit instructions are defined to ignore embedded instructions. (3) Capability inventory: The agent's capabilities include shell command execution and file writing. (4) Sanitization: The workflow incorporates brand-lint.mjs and video-gate for validation.
- Dynamic Execution: The skill references the execution of a build.mjs script to assemble final deliverables. Executing project-defined scripts is a standard development practice but involves the dynamic execution of code within the workspace.
Audit Metadata