general-video
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
hyperframesCLI vianpxfor project initialization, authentication, and status checks. It also executes local Node.js scripts for managing project preferences and applying recipes. - [EXTERNAL_DOWNLOADS]: Fetches visual treatments and components from a hosted registry using
npx hyperframes catalog. It also integrates with Figma to import design assets and tokens from external URLs. - [DYNAMIC_EXECUTION]: Employs a 'dispatch' pattern that generates instruction packets and worker roles to process scenes in parallel, allowing the agent to delegate sub-tasks within a controlled environment through generated instruction files.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: Processes user-provided
BRIEF.mdandSTORYBOARD.mdfiles, alongside external data from Figma URLs, to define the project scope and content. - Boundary markers: Utilizes a strict markdown format for storyboards and briefs to separate project metadata from content instructions, including the use of canonical terms defined in referenced contracts.
- Capability inventory: The agent uses file-system operations and shell command execution (
npx,node) to manage the video production pipeline. - Sanitization: The instructions explicitly require the use of specialized adapters for Figma assets to ensure proper sanitization of SVG content and verification of media provenance.
Audit Metadata