The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill employs dynamic module loading via import() with computed paths in scripts/lib/resolve.mjs and scripts/lib/clips.mjs. This mechanism is used to resolve dependencies from sibling directories within the ponti-studios toolset, relying on environment-variable-derived paths.\n- [COMMAND_EXECUTION]: The scripts invoke external binaries, specifically ffmpeg and ffprobe, through spawnSync to perform media processing tasks such as concatenation, duration analysis, and animatic generation.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes editorial data from scenes.json (ingestion point). The prompt and voiceover fields are interpolated into requests sent to external AI models (OpenRouter) without sanitization or protective boundary markers. The execution environment allows for file writing and network access, providing a surface for exploiting this injection.\n- [DATA_EXFILTRATION]: The skill performs network operations to the OpenRouter API for media generation and job polling. It also utilizes vendor utilities to publish keyframes to an external R2 storage service for provider access.

studio-mediagen