gemini-image-proxy_skill

Overview

This skill provides a lightweight Python wrapper to generate and edit images using Gemini 3 Pro Image via an OpenAI-compatible API endpoint and the openai Python SDK. It’s designed for simple, scriptable image generation and in-place edits with minimal dependencies. Install openai, set two environment variables, and run the included scripts to produce PNG/JPG/GIF/WEBP outputs.

How this skill works

The scripts call the openai SDK to interact with Gemini image models: client.images.generate(...) for new images and client.images.edits(...) for edits. You provide a text prompt and optional input file; the script sends the request to the configured OpenAI-compatible base URL using your API key and writes the returned image bytes to disk. It targets Python 3.10+ and requires only the openai package.

When to use it

• Create quick photorealistic or stylized images from text prompts in automation scripts.
• Apply edits to existing images (style changes, lighting, added elements) without a complex toolchain.
• Run on lightweight containers or platforms where minimizing dependencies is important.
• Prototype product shots, landscapes, or concept art from the command line.
• Integrate image generation into existing Python workflows using a single SDK.

Best practices

• Install the official openai package once (pip install openai) and pin a compatible version.
• Keep prompts concise but descriptive: include subject, style, lighting, and mood when relevant.
• Use the --input flag for targeted edits and preserve a copy of originals before overwriting.
• Set GOOGLE_PROXY_API_KEY and GOOGLE_PROXY_BASE_URL in environment variables rather than hardcoding keys.
• Test with preview models (if available) before scaling production requests to control cost.

Example use cases

• Generate marketing product images with a white background for e-commerce listings.
• Convert user photos into a specific art style (watercolor, comic, photorealistic).
• Add or remove simple scene elements (rainbow, sun, props) via edit prompts.
• Produce quick concept artwork for design reviews or storyboarding.
• Automate batch image generation for A/B testing visual variants.