The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill's core workflow requires the agent to write JavaScript files to lib/presets/ and subsequently execute them using node lib/render-cli.js. While this is the intended purpose, it represents a dynamic execution surface where agent-generated code is run locally.
[INDIRECT_PROMPT_INJECTION]: The skill identifies user-provided files in the ref/ directory (e.g., sketches, sample SVGs, or screenshots) as primary ingestion points for instructions. This creates a surface for indirect prompt injection if the processed files contain malicious instructions aimed at influencing the agent's code generation.
Ingestion points: User-provided reference files in ref/ (Step 1).
Boundary markers: None explicitly mentioned to distinguish between user intent and data content within those files.
Capability inventory: The skill has Write permissions to the filesystem and the ability to execute code via node (Step 4).
Sanitization: No explicit sanitization or validation of the contents of the reference files is defined before they are processed by the agent.

svg-animate