The Agent Skills Directory

[PROMPT_INJECTION]: The skill directly interpolates user-provided text—such as the dish name, kitchen style, and outfit description—into the prompts for both the image generation (Step 1) and video generation (Step 2). This allows for potential injection of instructions intended to bypass the underlying models' safety filters or alter the intended cinematic output.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which is then used to drive subsequent automated actions.
Ingestion points: Untrusted strings are accepted for dish, kitchen_style, and outfit parameters in SKILL.md.
Boundary markers: The prompt templates do not employ delimiters (like XML tags or triple quotes) or specific 'ignore' instructions to isolate user-provided data from the system instructions.
Capability inventory: The agent is instructed to execute shell commands via the muapi CLI and perform network operations using curl against the api.muapi.ai endpoint.
Sanitization: No sanitization or escaping mechanisms are described for the input variables before they are concatenated into the final CLI commands.
[COMMAND_EXECUTION]: The skill requires the agent to execute system commands (muapi image edit, muapi video generate, muapi predict wait) and network requests. While these are necessary for the skill's primary function, they provide an execution surface that relies on the presence of an API key (MUAPI_API_KEY) in the environment.

muapi-storyboard-to-cooking-video