The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill suggests installing the openai and pillow Python packages. These are standard, reputable libraries used for API interaction and image manipulation respectively.
[CREDENTIALS_UNSAFE]: The skill requires an OPENAI_API_KEY for live operations. The documentation follows security best practices by instructing the user to set this as a local environment variable and explicitly warns against sharing the full key in chat.
[COMMAND_EXECUTION]: The skill relies on executing a local Python script, scripts/image_gen.py. This script is a well-defined CLI tool that performs input validation on parameters like image size, quality, and output formats before making API calls.
[DATA_EXFILTRATION]: Data transfer is limited to sending prompts and images to the official OpenAI API endpoints. This is the intended functionality of the skill and is conducted using the official OpenAI SDK over secure HTTPS connections.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input through prompts and batch files in scripts/image_gen.py. It mitigates accidental instruction obedience by using structured prompt augmentation in the _augment_prompt_fields function, which adds clear labels like 'Primary request:' and 'Constraints:' to delineate user input from system instructions. The script's capabilities are appropriately restricted to image processing and API communication.

imagegen