The Agent Skills Directory

[SAFE]: The skill performs legitimate text generation tasks based on user-provided topics or transcripts. It follows a structured template for outputting JSON data intended for a design tool.
[DATA_EXFILTRATION]: The skill reads a configuration file located at ~/.claude/skills/carousel/config.json. This access is restricted to the skill's own application data directory for the purpose of retrieving user-defined settings, such as branding, voice, and character mappings, and does not constitute unauthorized data exposure.
[COMMAND_EXECUTION]: While the skill instructions tell the agent to ensure terminal commands included in the slides are 'real and correct,' the skill itself does not execute these commands or any arbitrary code on the host system.

carousel-copy