The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using subprocess.run to invoke ffmpeg and ffprobe for audio and video processing in scripts/convert_mp4.py and scripts/generate_tts.py.- [COMMAND_EXECUTION]: The SKILL.md file instructs the agent to use the afplay utility to play generated audio files locally on macOS.- [CREDENTIALS_UNSAFE]: The scripts/upload_youtube.py script automatically searches for and reads Google OAuth client secrets from standard user directories, including ~/Downloads/ and ~/.config/google/.- [CREDENTIALS_UNSAFE]: The skill's instructions in SKILL.md describe passing the OpenAI API key as a command-line argument to generate_tts.py, which may expose the key in system process logs.- [EXTERNAL_DOWNLOADS]: Fetches audio data from OpenAI's TTS API and uploads video content to Google's YouTube Data API.- [PROMPT_INJECTION]: The skill processes untrusted content from external URLs, social media posts, and repositories, creating a surface for indirect prompt injection during script generation.
Ingestion points: Fetches data from URLs, Twitter/X posts (via api.fxtwitter.com), PDFs, and GitHub repositories as defined in SKILL.md.
Boundary markers: The instructions lack explicit delimiters or instructions to ignore embedded commands in the ingested source text.
Capability inventory: The skill can perform network operations, write to the local file system, and execute shell commands via Python scripts.
Sanitization: The scripts/generate_tts.py script removes markdown formatting and tags from the script but does not validate or sanitize for embedded natural language instructions.

podcast