The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructions provide patterns for using new Function() to execute dynamically assembled JavaScript code strings within the Figma plugin sandbox.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from component contract JSON files to drive its generation logic and script assembly.
Ingestion points: Configuration and component definitions are read from .component-contracts and contract files in the {CONTRACTS_DIR}.
Boundary markers: The instructions do not define boundary markers or safety warnings for the data ingested from contracts.
Capability inventory: The skill executes code in the Figma environment via use_figma, writes temporary files to the scripts/ directory, and runs shell commands.
Sanitization: There are no explicit sanitization or validation steps specified for the contract data before it is incorporated into scripts.
[COMMAND_EXECUTION]: The skill uses shell commands, including npx rimraf and rm, for managing temporary working files and performing post-run cleanup.
[EXTERNAL_DOWNLOADS]: The skill utilizes npx rimraf, which involves downloading the utility from a well-known package registry if it is not locally available.

cc-figma-component