The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to use npx -y @google/design.md to lint and export design tokens. This involves downloading and executing a package from a well-known organization's registry scope.
[INDIRECT_PROMPT_INJECTION]: The skill processes DESIGN.md files which contain a mix of YAML tokens and Markdown prose, creating a surface for potential injection attacks.
Ingestion points: The agent ingests DESIGN.md files provided by the user or read from the local file system during linting, diffing, and export workflows.
Boundary markers: There are no explicit instructions or delimiters defined to separate the data tokens from potential instructions embedded within the Markdown body of the files.
Capability inventory: The skill uses write_file to save specifications and executes shell commands (npx) to process the files.
Sanitization: The provided lint command validates token structures and WCAG contrast but does not sanitize the human-readable Markdown prose for hidden prompt instructions.

design-md