The Agent Skills Directory

[SAFE]: No malicious behavior detected. The skill implements a creative media workflow using legitimate third-party tools and vendor-owned resources.\n- [COMMAND_EXECUTION]: The skill executes ffmpeg and ffprobe to manipulate media files according to a generated cut manifest. These operations are restricted to the local workspace and intended media files.\n- [EXTERNAL_DOWNLOADS]: The workflow involves installing several specialized media generation skills (e.g., whisperx, p-video-avatar) from the PrunaAI repository. These references are consistent with the vendor's provided ecosystem.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied lyrics and visual prompts. 1. Ingestion points: music-video-plan.template.json. 2. Boundary markers: Section tags (e.g., [Verse], [Chorus]). 3. Capability inventory: ffmpeg, ffprobe, and curl network operations in SKILL.md. 4. Sanitization: Absent, relies on downstream model safety protocols.

music-video