The Agent Skills Directory

[SAFE]: The skill features robust defensive measures against common SVG-based attacks. It includes a dedicated validation script (scripts/validate_svg.py) that identifies and prevents the use of malicious scripting (event handlers, javascript: URLs), dangerous XML constructs (XXE), and unsafe external resource fetching.
[COMMAND_EXECUTION]: The skill uses the Bash(python3:*) tool to run its internal validation script. This is a legitimate and secure implementation designed to ensure the technical correctness and safety of generated SVG files before delivery.
[PROMPT_INJECTION]: Although the skill is designed to repair and optimize user-provided SVG markup (an indirect prompt injection surface), the instructions explicitly mandate strict sanitization and validation of this data to prevent the agent from executing instructions embedded within the SVG code.

svg-creator