The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's workflow depends on the execution of several local scripts and command-line utilities. This includes a shell script for asset collection (collect_assets.sh), a Node.js script for browser-based screen recording (record_demo.js), and a Python script for voiceover generation (generate_voiceover.py). It also utilizes FFmpeg for audio/video manipulation and the Remotion framework for video synthesis.
[EXTERNAL_DOWNLOADS]: The skill incorporates external content from well-known stock media providers including Unsplash, Pexels, and Pixabay for image assets. It also leverages the edge-tts service and mentions Azure TTS as a high-quality alternative for voice synthesis.
[INDIRECT_PROMPT_INJECTION]: The skill demonstrates a vulnerability surface for indirect prompt injection as it processes untrusted user input and external assets through a high-capability execution environment.
Ingestion points: User-provided product themes, descriptions, and externally sourced image/video assets (Phase 2).
Boundary markers: None identified in the skill instructions to distinguish between instructions and data.
Capability inventory: Significant capability including shell script execution, headless browser automation (Playwright), and local file system operations for video rendering.
Sanitization: No explicit sanitization or validation steps are defined before passing user-provided strings to the TTS engine or the rendering scripts.

product-video