The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided content such as topics, brand information, and creative briefs which could potentially contain malicious instructions. However, the risk is mitigated by the specific domain of motion design and standard LLM guardrails.
Ingestion points: User-provided topics and brand facts mentioned in SKILL.md.
Boundary markers: The instructions do not explicitly mandate the use of delimiters for all user-provided inputs.
Capability inventory: The skill can trigger tool execution via motion-production and read local reference files.
Sanitization: No specific sanitization or validation of user-provided content is mentioned.
[DYNAMIC_EXECUTION]: The skill involves the generation of programmatically controlled visual elements like HTML, SVG, or 3D assets. The agent generates code or scripts that are later executed by a motion rendering engine.
Evidence: SKILL.md and production-decisions.md describe the creation of 'editable templates' and 'programmatic materials'. This is a core feature of the tool and is handled through established templates.
[EXTERNAL_DOWNLOADS]: The skill references documentation and design principles from external repositories.
Evidence: references/examples.md links to official GitHub repositories for Remotion, Heygen, and iart-ai for methodology reference. These are well-known technology organizations and the references are used for informational purposes.

motion-director