The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/slide-token-validator.py uses subprocess.run to delegate HTML validation to a secondary local script scripts/html-token-validator.py. The execution uses a list-based command structure without shell=True, which prevents shell injection and restricts execution to the intended local file.
[EXTERNAL_DOWNLOADS]: The skill references external assets from well-known and trusted services. Specifically, scripts/fetch-background.py contains hardcoded image URLs from Pexels for use as slide backgrounds, and the slide generator in scripts/generate-slide.py includes a reference to the Chart.js library hosted on the jsdelivr CDN. These are standard practices for web-based slide generation.
[DATA_EXPOSURE]: No hardcoded credentials or sensitive file paths were detected. The scripts process local design token configurations (JSON) and layout databases (CSV). The skill-report.json file contains references to 'hex color codes' being flagged as weak crypto, but these are confirmed to be benign CSS color values (e.g., #2563EB).
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local CSV files to populate slide content and strategies. While this constitutes an attack surface, the data sources are provided as part of the skill package rather than being fetched from untrusted runtime sources. The logic for interpolating this data into HTML is consistent with its stated purpose as a slide generator.

ckm-design-system