The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md reference the execution of local Python scripts (scripts/design_lint.py and scripts/run_design_evals.py) to perform design validation. These are localized to the skill's own environment for utility purposes.
[SAFE]: The Python scripts perform standard string processing and JSON validation (linting) without network access, obfuscation, or sensitive data handling. The logic is focused on detecting 'vague' design adjectives and 'slop' visual effects.
[SAFE]: JSON schemas and assets (SVG logos) contain standard structured data and graphics without hidden payloads.

composition-director