The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's core workflow involves generating and executing complex shell commands (specifically FFmpeg and file utilities) based on parameters derived from user-provided or data-sourced text files.
Evidence: In workflows/02_draft_processing.md, the instructions state: 'Execute all generated FFmpeg instructions in order.' Additionally, workflows/01_asset_organization.md uses the stat command to gather file metadata.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests instructions from external sidecar documents to determine the parameters for command execution, without explicit sanitization or boundary markers.
Ingestion points: workflows/02_draft_processing.md (reading .txt sidecar files containing editing intent).
Boundary markers: Absent. No specific delimiters or warnings to ignore embedded instructions are defined for the sidecar file parsing.
Capability inventory: Subprocess execution via shell for ffmpeg, stat, and file renaming operations.
Sanitization: Absent. The skill instructions do not specify any validation, escaping, or filtering of the 'editing intent' strings before they are interpolated into FFmpeg command templates.

showcase-video-processor