The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a measurement utility in scripts/measure.py that the agent executes to verify image properties. This script uses subprocess.run() to call a system browser (Chrome or Edge) in headless mode to rasterize SVG files. The implementation follows security best practices by passing arguments as a list rather than a shell string, which prevents shell injection.
[EXTERNAL_DOWNLOADS]: The skill fetches illustration assets (SVGs and PNGs) from the Icons8 catalog (icons8.com) via tool calls. These downloads are legitimate and necessary for the skill's primary function of shipping design assets.
[INDIRECT_PROMPT_INJECTION]: The skill processes external SVG content which could theoretically contain malicious instructions or browser exploits.
Ingestion points: scripts/measure.py reads SVG files downloaded to the local project directory via tool calls.
Boundary markers: None present in the data processing flow.
Capability inventory: The measure.py script utilizes subprocess.run to spawn browser processes and uses the pillow library for image analysis.
Sanitization: No explicit sanitization of SVG content is performed before rendering.
Assessment: The risk is considered minimal as it is inherent to the functional requirement of analyzing vector artwork.
[SAFE]: All identified behaviors, including the use of local scripts and external asset fetching, are consistent with the skill's stated purpose and originate from a recognized vendor (Icons8).

ouch