The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute specific Python scripts (build_style_brief.py, validate_style_brief.py, and score_aesthetics.py) distributed with the skill. These scripts are used for configuration generation and aesthetic scoring and do not accept arbitrary shell commands.
[DATA_EXPOSURE]: File operations are restricted to a project-relative .clawdcut directory. The scripts read project files (script.md, storyboard.md, style_brief.json) and write reports (aesthetic_score.json, aesthetic_report.md). There is no access to sensitive system directories or credentials.
[INDIRECT_PROMPT_INJECTION]: The aesthetic scoring script processes untrusted data from user-provided scripts and storyboards. However, the ingestion is safe as the script only performs keyword matching to calculate a numerical score and does not execute the file content or interpolate it into critical instructions.
[SAFE]: The Remotion code snippets provided in the rules/ directory are for documentation and user guidance. They demonstrate standard UI and animation logic using React and do not contain hidden malicious payloads or network calls.

video-aesthetics