The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of external image and video URLs which are then processed by the HeyGen API.
Ingestion points: source_image_url and target_video_url parameters defined in SKILL.md.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore potential malicious content embedded in the processed media metadata or source files.
Capability inventory: The skill performs network requests to the HeyGen API using the requests library in Python and fetch in TypeScript.
Sanitization: There is no evidence of URL validation or content sanitization for the inputs before they are transmitted to the remote service.

faceswap