The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/vtd.js executes external CLI tools based on user-supplied arguments. The run function utilizes node:child_process.spawn to execute yt-dlp and ffmpeg. The script includes an explicit mechanism to pass arbitrary extra arguments to yt-dlp using the -- separator, which grants access to advanced tool features like command execution through specific flags.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external URLs, creating a surface for indirect prompt injection attacks where malicious instructions could be embedded in video transcripts or subtitles.\n
Ingestion points: Data is fetched from external URLs provided to the transcript, download, audio, and subs commands in scripts/vtd.js.\n
Boundary markers: The output provided to the agent is not wrapped in security delimiters or provided with warnings to ignore embedded instructions.\n
Capability inventory: The skill has the ability to execute subprocesses (yt-dlp, ffmpeg) and perform file system writes to the user's ~/Downloads directory and system temporary directories.\n
Sanitization: The cleanSegments function in scripts/vtd.js strips HTML tags, timestamps, and music cues to improve readability, but it does not filter or sanitize the content for potential natural language instructions.

video-transcript-downloader