website-to-hyperframes
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to capture and process untrusted data from external websites (Step 1) to generate creative artifacts like scripts, storyboards, and design references. This creates a vulnerability surface where a website's content could attempt to influence or override the agent's behavior during the creative process.
- Ingestion points: Website capture data processed in Step 1.
- Boundary markers: Absent; the instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the captured content.
- Capability inventory: The skill writes multiple project files (SKILL.md mentions writing DESIGN.md, SCRIPT.md, and STORYBOARD.md) and executes CLI tools (npx hyperframes).
- Sanitization: Absent; there is no mention of sanitizing or filtering the captured text before it is used to generate new content.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute command-line tools using
npx(specificallynpx hyperframes lintandnpx hyperframes validate) to verify the project status. While these are ecosystem-specific tools, they involve the execution of external packages.
Audit Metadata