The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it interpolates user inputs directly into prompt templates for image and video generation.
Ingestion points: The baby_costume, dialogue, and baby_description inputs in SKILL.md.
Boundary markers: None present to distinguish instructions from user data.
Capability inventory: Calls muapi CLI tools and curl.
Sanitization: No evidence of input filtering or escaping.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands using the muapi CLI and curl, which could lead to command injection if the user-provided inputs are not handled safely by the underlying shell or tools.
[DATA_EXFILTRATION]: The skill initiates network connections to api.muapi.ai to transmit user content and an API key. While functional for the skill's purpose, this involves sending data to a domain not on the standard whitelist.

muapi-talking-baby-video