The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing untrusted data through the Kling video generation models.
Ingestion points: External data enters the context via the image_url parameter and user-supplied prompt inputs for the Kling models.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the provided templates.
Capability inventory: The skill facilitates network-based API calls to the fal.ai platform to process video generation requests.
Sanitization: There are no explicit sanitization or validation mechanisms mentioned for external content prior to model interpolation.
[EXTERNAL_DOWNLOADS]: The skill downloads and processes image data from external URLs specified in the image_url parameter. While necessary for the stated functionality, this involves fetching data from remote sources.
[DATA_EXFILTRATION]: The skill provides instructions for managing secrets via the FAL_KEY environment variable. This is documented as standard secret management practice for authenticating with the fal.ai service.

kling