The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/channel_performance.py script executes the yt-dlp command-line utility to retrieve video metadata. It uses the subprocess.run() function with a list of arguments rather than a shell string, which is a secure practice that prevents shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill relies on the third-party utility yt-dlp to function. The script provides instructions for the user to install this dependency manually via standard package managers.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it processes untrusted data (YouTube video titles) from external sources.
Ingestion points: External titles are fetched and printed by scripts/channel_performance.py for the agent to analyze.
Boundary markers: The output uses simple text dividers but lacks specific instructions to the agent to ignore embedded commands within the fetched titles.
Capability inventory: The skill's primary capability is text generation (brainstorming titles).
Sanitization: There is no filtering or escaping of the ingested video titles before they are presented to the agent. However, given the skill's specific and non-critical purpose, this poses a very low risk.

yt-titles