The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run a basic ls command to verify the presence of its 'kit' directory structure (tokens folder). This is a benign diagnostic check used to ensure the skill has access to necessary design tokens before proceeding with optimization tasks.
[DATA_EXPOSURE]: The skill uses relative path traversal (../../../) to access files outside its immediate directory. While this pattern can be risky, the files targeted (sizing.json, motion.json, and performance.md) are non-sensitive design tokens and workflow documentation belonging to the vendor's 'kit' infrastructure. No sensitive system files or credentials are accessed.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external files (performance.md, sizing.json) which influences its diagnostic and optimization behavior. Although this creates an ingestion surface for external data, the resources are local vendor-provided configuration files rather than untrusted third-party inputs.

performance