The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the npx hyperframes CLI tool and local Node.js scripts (e.g., frame-packets.mjs, prefs.mjs, recipe.mjs) to manage project initialization, authentication status, and the rendering pipeline.
[EXTERNAL_DOWNLOADS]: The skill fetches assets and design tokens from figma.com URLs and uses npx to execute tools, which may involve downloading packages from the registry.
[DYNAMIC_EXECUTION]: The skill dynamically generates sub-agent role definitions (_role.md) by concatenating the core framework contract with skill-specific instructions at runtime before dispatching scene-building tasks.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources and local project files to drive the video generation process.
Ingestion points: Project configuration files (BRIEF.md, STORYBOARD.md) and assets from figma.com URLs.
Boundary markers: The workflow relies on the brief-contract.md specification and canonical term enforcement to maintain structure, though explicit prompt delimiters for external content are handled by the underlying frame-packets-core.mjs library.
Capability inventory: File system write operations for HTML and JSON artifacts, shell command execution via the HyperFrames toolchain, and sub-agent task delegation.
Sanitization: Implements mandatory validation phases using npx hyperframes lint and npx hyperframes check to ensure generated compositions adhere to structural and deterministic layout rules.

general-video