The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions guide the agent to construct and run a shell command using parameters directly derived from user input without specifying requirements for sanitization.
Ingestion points: User-provided image descriptions and output filenames as defined in the Step 1 and Step 2 workflows in SKILL.md.
Boundary markers: The provided command template (--prompt "your prompt here") uses simple double quotes that do not provide adequate protection against shell metacharacters if the agent performs literal string interpolation.
Capability inventory: The skill facilitates command execution of a local Python script (scripts/generate_image.py) and has the capability to write files to the system via the script's save_image_bytes function.
Sanitization: The skill documentation lacks any instructions for the agent to validate, escape, or sanitize the user-provided data before passing it to the command line interface.
[COMMAND_EXECUTION]: The skill's primary fallback mechanism involves executing a local Python script using the system shell, which serves as the vector for the potential command injection vulnerability.

image-generation