The Agent Skills Directory

[DATA_EXFILTRATION]: The skill transmits user-provided data (prompts and image URLs) to the external domain three.ws for processing. While this is the primary functionality, it involves network operations to a non-whitelisted domain.
[COMMAND_EXECUTION]: The skill documentation includes examples of curl commands intended for manual interaction with the three.ws REST API, which involves executing shell commands to perform network operations.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input to drive 3D generation, creating a potential attack surface.
Ingestion points: The prompt and image_url arguments in the forge_avatar tool and manual API calls (SKILL.md).
Boundary markers: Absent. There are no instructions to delimit user input or warn the agent about embedded instructions in the prompt data.
Capability inventory: The skill has the ability to perform network requests and receive external URLs via the forge_avatar tool and curl examples (SKILL.md).
Sanitization: Absent. The skill does not describe any validation or filtering of user-supplied prompts or image references.

create-3d-avatar