The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection.
Ingestion points: The --text and --url flags in the present and picstory commands (SKILL.md) allow ingestion of untrusted content into the AI script generation pipeline.
Boundary markers: The instructions do not specify the use of delimiters or system-level instructions to ignore malicious content within the provided text or URL data.
Capability inventory: The skill utilizes the voxflow CLI for network communication and ffmpeg for local file system operations (SKILL.md).
Sanitization: There is no mention of sanitizing or validating the content retrieved from external URLs or input text before it is processed by the LLM.
[COMMAND_EXECUTION]: The skill relies on local execution of command-line tools for media processing and service interaction.
Evidence: Use of ffmpeg for video rendering and the voxflow CLI for authentication, status checks, and content generation is documented in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of external dependencies and software.
Evidence: Instructions include installing the voxflow global Node.js package and system-level ffmpeg (SKILL.md).

video