The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run across several scripts (render_cut.py, gen_shorts.py, gen_edl.py) to invoke ffmpeg and ffprobe for video processing. These operations are essential to the skill's functionality. The implementation correctly uses list-based arguments for subprocess calls, which is a security best practice that prevents shell injection vulnerabilities.- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation and use of the faster-whisper library, which involves downloading AI models (e.g., large-v3) from official repositories upon first use. It also references standard installation procedures for well-known tools like ffmpeg and Python. These downloads are from trusted sources and are consistent with the skill's purpose.- [PROMPT_INJECTION]: The skill implements a workflow where the agent reads and reviews transcripts generated from user-provided video files (scan_terms.py). While this represents a surface for indirect prompt injection (where instructions hidden in audio could influence the agent), the skill includes manual review gates (G2, G6) and specific instructions for the agent to use its own judgment, which mitigates the risk.- [CREDENTIALS_UNSAFE]: The skill maintains safe credential practices by explicitly instructing users to store sensitive keys (like optional ElevenLabs API keys) in environment variables rather than hardcoding them or writing them to disk. No hardcoded credentials were found in the provided files.

youtube-edit-kit