The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines shell commands for avifenc and exiftool that incorporate user-provided variables for file paths and settings. * Evidence: The procedure section in SKILL.md contains shell templates like avifenc -q <quality> -s <speed> [--lossless] [--depth <n>] "<input>" "<output>". * Risk: If the execution environment does not properly sanitize these variables, specifically file paths or options, it could lead to command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes external image files which can contain embedded instructions designed to influence the behavior of the AI agent. * Ingestion points: User-provided JPEG, PNG, and TIFF files processed by the skill. * Boundary markers: No explicit boundary markers or instructions to ignore embedded content are provided. * Capability inventory: Shell execution capabilities using avifenc and exiftool. * Sanitization: No sanitization or validation of the image content or metadata is performed before or after processing.

convert-to-avif