The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill interpolates the user-provided subject_description directly into prompts for image generation and editing tasks without using boundary markers or sanitization. This creates an attack surface where malicious input could influence the generated output or agent behavior.
Ingestion points: The subject_description input field in SKILL.md.
Boundary markers: None are used to delimit the user input within the prompts.
Capability inventory: The skill uses muapi CLI commands and curl for network requests to api.muapi.ai.
Sanitization: No input validation or filtering is performed on the description before it is processed.
[COMMAND_EXECUTION]: The skill instructs the agent to execute muapi CLI commands (muapi image generate, muapi image edit, muapi auth configure) and curl commands to interact with the external MuAPI service. These commands are necessary for the skill's primary functionality and utilize environment variables for authentication.

muapi-multi-angle-reshoot