{'>'}_
English 繁體中文 日本語 한국어 Deutsch Español Français
Login
0 stars
0 forks
Python
38 views
View on GitHub Add to Favorites

SKILL.md

Image Generation & Editing Skill (Azure OpenAI)

This skill allows the agent to generate new images from text prompts or edit existing images using reference files via Azure OpenAI's GPT-Image-1.5 model.

Configuration

  • Script: gen.py
  • Environment: /home/thomas/.env.azure (Required for API credentials)
  • Output: Images are saved to the ./out/ directory with a timestamped filename.

Usage

1. Generate New Image

Use this mode when the user wants to create an image from scratch. Syntax: python3 gen.py "<detailed_prompt>"

2. Edit Existing Image

Use this mode when a user provides one or more reference images and requests changes (e.g., changing colors, adding objects, or altering style). Syntax: python3 gen.py "<edit_instructions>" --edit --ref <path_to_image>

Parameters

Parameter Type Description
prompt String (Positional) The visual description or edit instruction.
--edit Flag Required to trigger image-to-image/editing mode.
--ref Path(s) One or more paths to the images to be used as a reference.

Examples

  • Create a new image: python3 gen.py "A professional headshot of a solution consultant in a modern office, cinematic lighting"

  • Modify an existing image: python3 gen.py "Change the suit color to charcoal gray and add a laptop on the desk" --edit --ref ./out/20260208_120000.png

Notes for the Agent

  • Fidelity: The skill is hardcoded to "high" fidelity to ensure identity and structural persistence during edits.
  • Image Formats: Input images must be in PNG format.
  • Output: The tool returns a JSON object containing the path to the generated image. Always display this path or the image to the user.