The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface by processing natural language video briefs and user arguments to generate narrative plans and React code for video scenes.
Ingestion points: Processes user-provided video briefs and arguments (referenced as $ARGUMENTS in SKILL.md) to define scene content and structure.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" warnings for the user-supplied content.
Capability inventory: The skill has the capability to execute shell commands via npx remotion still and npx remotion render across various scripts (referenced in SKILL.md and rules/scene-evaluator.md).
Sanitization: No explicit sanitization, validation, or escaping of the user-provided text is described before it is interpolated into code templates.
[COMMAND_EXECUTION]: The skill utilizes standard framework commands to perform its core functions.
Evidence: Executes npx remotion still in Phase 4b to capture scene previews and npx remotion render in Phase 5 to generate final video files.
[EXTERNAL_DOWNLOADS]: The skill references and uses well-known industry libraries and tools.
Packages: Dependencies include remotion, gsap, and @remotion/transitions, which are standard for web-based animation and video production.
Note: These resources are recognized as part of the intended functionality and originate from well-known sources.
[DYNAMIC_EXECUTION]: The skill involves the runtime generation and execution of code as its primary purpose.
Process: React components and configuration files are generated dynamically from narrative templates and then executed by the Remotion engine during preview and rendering phases.

video-producer