The Agent Skills Directory

[PROMPT_INJECTION]: Potential for indirect prompt injection through external video data.
Ingestion points: Transcripts are downloaded from YouTube and read into the agent's context via Python scripts in SKILL.md.
Boundary markers: Absent. The skill does not utilize delimiters or specific instructions to the agent to treat the transcript content as untrusted data.
Capability inventory: The skill has access to the Bash tool, allowing for command execution and file system interaction.
Sanitization: The scripts perform HTML unescaping for readability but do not filter or sanitize the content for malicious prompt instructions.
[COMMAND_EXECUTION]: Utilizes the Bash tool to execute yt-dlp for metadata retrieval and python for parsing transcript files.
[EXTERNAL_DOWNLOADS]: Downloads the yt-dlp tool and video transcript data from external servers during execution.

youtube