The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it interpolates user-controlled data directly into tool prompts.
Ingestion points: The person_description and product_image inputs are accepted from the user in SKILL.md.
Boundary markers: There are no delimiters or instructions to the model to ignore embedded commands within the user-provided description.
Capability inventory: The skill can execute shell commands via the muapi CLI and perform network operations using curl.
Sanitization: No escaping, validation, or filtering is applied to the user input before it is used in Phase A or Phase B prompts.
[COMMAND_EXECUTION]: The skill instructs the agent to execute muapi CLI commands (e.g., muapi image edit, muapi video from-image, muapi predict wait) and curl commands to interact with the vendor's API.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to api.muapi.ai to facilitate image and video generation tasks. These requests utilize the MUAPI_API_KEY environment variable, which is a standard security practice for managing credentials.

muapi-giant-product-showcase