The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill takes multiple user inputs—dish, kitchen_style, and outfit—and interpolates them directly into the prompts for image and video generation tools.
Ingestion points: Inputs are defined in the Inputs section of SKILL.md and used in Step 1 and Step 2.
Boundary markers: The prompt templates do not use delimiters (like triple quotes or XML tags) or specific instructions to the underlying model to ignore instructions contained within those user fields.
Capability inventory: The skill uses the muapi CLI to interact with external AI services, requiring network access and environment variables for authentication.
Sanitization: There is no evidence of input validation or sanitization to prevent a user from providing malicious instructions that could influence the AI's output beyond the intended scope.
[COMMAND_EXECUTION]: The skill relies on executing the muapi CLI tool and potentially curl to interact with the vendor's API.
The instructions explicitly direct the executing agent to run shell commands to generate media.
Authentication involves managing the MUAPI_API_KEY via environment variables or CLI configuration.

muapi-storyboard-to-cooking-video