The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script hookscore.py to analyze hook options. The script uses regular expressions defined in a local hooks.json file to evaluate text for engagement metrics like specificity and curiosity.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external data from user-provided video transcripts to infer a "voice profile," creating a vulnerability to instructions embedded in that data.
Ingestion points: Processes raw ideas and video content (transcripts) to generate writing styles and scripts as instructed in SKILL.md.
Boundary markers: The skill lacks explicit markers or instructions to isolate potentially adversarial content within the ingested transcripts.
Capability inventory: The agent is authorized to write to ~/.claude/youtube/voice.md and execute local Python commands for text analysis.
Sanitization: There is no evidence of sanitization or filtering applied to the external transcript content before it is used to define the user's voice profile or generate script content.

yt-script