The Agent Skills Directory

EXTERNAL_DOWNLOADS (MEDIUM): The script scripts/download_video.py automatically installs the yt-dlp package using pip at runtime if it's not detected. This bypasses static dependency checks and introduces a risk of installing malicious packages if the environment or registry is compromised.
COMMAND_EXECUTION (MEDIUM): The skill executes external commands using subprocess.run. Although it passes arguments as a list (mitigating shell injection), it still runs a complex external tool (yt-dlp) on untrusted input (URLs). This could potentially expose the system to vulnerabilities within yt-dlp or its dependencies.
INDIRECT_PROMPT_INJECTION (LOW): The skill ingests untrusted data and passes it to an external process.
- Ingestion points: url argument in scripts/download_video.py.
- Boundary markers: None (the URL is used directly).
- Capability inventory: subprocess.run to execute system commands.
- Sanitization: Uses argument lists for subprocess.run, which prevents direct shell injection but does not sanitize the URL content processed by yt-dlp.
PROMPT_INJECTION (SAFE): No evidence of malicious prompt overrides or instructions to bypass safety filters were found in the provided files.

youtube-downloader