The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a shell script template for fetching node geometry data from an external source.
The script utilizes curl to access api.figma.com, which is a well-known and trusted service for this domain.
It implements protective measures such as umask 077 to restrict file permissions and mktemp for secure temporary file creation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from the Figma API, which represents a potential ingestion surface.
Ingestion points: Data is fetched from the Figma REST API via the /v1/files/{key}/nodes endpoint.
Boundary markers: The skill includes instructions to validate the JSON response and maintain a clear separation between implementation assumptions and fetched values.
Capability inventory: Capabilities include network access via curl and local file system writes for caching API responses.
Sanitization: The skill mandates validating that the fetched content is valid JSON. The nature of the data (primarily numeric geometric properties) significantly limits the risk of adversarial instruction execution.

figma-raw-geometry