The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user inputs such as brand_name, industry, and style_direction and interpolates them directly into image generation prompts.
Ingestion points: Inputs defined in SKILL.md are directly substituted into template strings in the generation phases.
Boundary markers: The skill lacks any delimiters, quotes, or "ignore embedded instructions" warnings to isolate user-provided data from the agent's instructions.
Capability inventory: The skill possesses the ability to execute shell commands (muapi CLI) and perform network operations (curl) across all defined generation steps.
Sanitization: There is no evidence of validation, escaping, or filtering of the external content before it is interpolated into the prompts.
[COMMAND_EXECUTION]: The skill instructions direct the agent to perform shell operations, including configuring authentication via muapi auth configure and making HTTP requests using curl to external endpoints.
[DATA_EXFILTRATION]: The skill accesses the sensitive MUAPI_API_KEY environment variable and transmits it in the headers of curl requests to the external domain api.muapi.ai. While this is intended for API functionality, it represents the transmission of credentials to a non-whitelisted third-party service.

muapi-design-guide