gemini-api

The Gemini API skill provides a streamlined interface for interacting with Google's advanced image generation models, specifically optimized for high-quality production workflows. It enables developers to generate, manipulate, and iterate on visual assets directly through a command-line interface. By integrating support for models such as gemini-3-pro-image-preview and gemini-2.5-flash-image, this skill balances high-fidelity 4K output with rapid iterative prototyping. It is designed for engineers and designers who require programmatic control over image generation without manual overhead, ensuring that visual content creation becomes a scalable part of their software delivery pipeline.

• Support for text-to-image generation with customizable aspect ratios ranging from social media standard 1:1 to cinematic 21:9.

• Native capabilities for image editing and style application using external reference images to maintain visual consistency.

• Advanced CLI flagging system, allowing fine-grained control over model IDs, retry strategies, and style templates.

• Automated exponential backoff and retry mechanism to handle transient network errors, rate limiting (429), and server-side timeouts.

• Comprehensive error handling categorized for quick recovery from API key issues, input validation errors, and content policy restrictions.

• Users must provide a valid GEMINI_API_KEY as an environment variable to authenticate requests.

• For consistent stylistic outputs, utilize the --style flag with markdown-based templates.

• The tool is best used in combination with batch processing to generate multiple variations of a prompt for iterative refinement.

• Be aware that content policy violations will trigger specific error codes, requiring prompt adjustments.

• Monitor the usage of retry logic, which defaults to 3 attempts, and adjust the --max-retries flag for high-latency or unstable network environments.

• Ideal for automated content generation, UI asset creation, and rapid prototyping of marketing materials within CI/CD pipelines.