The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided scripts and image descriptions that are sent to external AI models for video generation, creating a surface for potential prompt injection.
Ingestion points: The input_text field in references/video-generation.md and the appearance field in references/photo-avatars.md accept arbitrary user strings.
Boundary markers: No delimiters or protective instructions are demonstrated in the provided request examples to prevent the model from following commands embedded in user text.
Capability inventory: The skill demonstrates network communication with heygen.com and remotion rendering capabilities.
Sanitization: The reference code does not include sanitization or validation logic for script inputs.
[SAFE]: The skill targets legitimate, well-known domains for its operations. Sensitive credentials like HEYGEN_API_KEY are retrieved from environment variables rather than being hardcoded, which aligns with security best practices. Automated scanner alerts on the SKILL.md file were evaluated and determined to be false positives based on common coding patterns.

avatar-video