The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external project files, such as token definitions and source code, to generate mapping reports.
Ingestion points: Token files, theme entries, and product source code files (SKILL.md).
Boundary markers: None explicitly mentioned to delimit or ignore instructions within processed files.
Capability inventory: Restricted to reading files, executing git status, and running the oklch.mjs utility script (SKILL.md, references/mapping-rules.md).
Sanitization: The skill normalizes values (e.g., converting colors to OKLCH) using a script before processing, which serves as a validation step for style data.
[COMMAND_EXECUTION]: The skill executes the git status command to ensure no files are modified and runs a local vendor-provided script build-design-system/scripts/oklch.mjs to perform color space conversions.

token-mapping