The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run across several core scripts (runner.py, utils.py, single_task.py) to manage the different phases of the image generation pipeline. These calls facilitate the execution of internal tasks and the invocation of sibling AIGC tools within a controlled environment.
[DYNAMIC_EXECUTION]: The logic in single_task.py employs importlib.import_module to load internal parameter-building modules dynamically. This is used for modular code organization, with module names resolved from the skill's own script directory.
[EXTERNAL_DOWNLOADS]: Includes functionality to download user-provided apparel images for processing. This is implemented using standard library functions (urllib.request) and targets user-supplied assets required for the skill's primary function.
[PROMPT_INJECTION]: The skill processes user-supplied text (selling points and scene descriptions) and interpolates them into prompt templates. To mitigate potential injection risks, it automatically appends a 'Sensitivity Suffix' to prompts, instructing the model to avoid infringing or restricted terminology.
[DATA_EXFILTRATION]: Analysis did not reveal any unauthorized data harvesting or transmission. Network activity is restricted to functional requirements, such as downloading reference images and communicating with the vendor's designated tool gateway.

linkfox-aigc-imagegen-cloth