nano-banana-pro

What This Skill Does

Nano Banana Pro is a specialized agent skill designed to bridge OpenClaw with the Gemini 3 Pro Image generation engine. It allows users to generate, modify, and compose visual assets directly from the CLI or chat interface. By leveraging a streamlined Python-based integration, it handles image creation tasks ranging from simple text-to-image prompts to complex multi-image visual composition. The skill automatically manages output handling and reports media paths back to the agent for seamless file integration.

Installation

To install this skill, use the ClawHub CLI command in your terminal: clawhub install openclaw/skills/skills/autogame-17/nano-banana-pro-multi-image

Ensure you have your Gemini API key configured. You can set it as an environment variable GEMINI_API_KEY or define it within your ~/.clawdbot/openclaw.json file under skills.nano-banana-pro.apiKey.

Use Cases

• Rapid Prototyping: Quickly generate visual concepts or UI mockups during the development process.
• Asset Composition: Merge multiple source images into a single cohesive scene for artistic or technical storytelling.
• Image Refinement: Use existing image files as references or starting points to apply stylistic changes or structural edits via natural language instructions.
• Automated Reporting: Generate visual charts, diagrams, or artistic headers automatically for documentation workflows.

Example Prompts

1. "Generate a high-resolution 2K image of a futuristic cyberpunk city with neon lights and flying cars, save it as 2023-10-27-14-00-00-cyber-city.png."
2. "Edit the file logo.png to change the color scheme to autumn tones and increase the sharpness to 4K resolution."
3. "Combine these three sketches: img1.png, img2.png, and img3.png into a single architectural concept render."

Tips & Limitations

• Filename Formatting: Always use the recommended yyyy-mm-dd-hh-mm-ss-name.png format to keep your assets organized and avoid overwriting existing files.
• Resolution Scaling: While 1K is the default, remember that 4K generation consumes more computational resources and API credits. Choose based on your specific output requirements.
• Composition Limits: You can compose up to 14 images at once; however, ensuring your source images are consistent in lighting and perspective will yield better results.
• Resource Management: Do not attempt to read the image output files via the agent's reading functions; the script outputs the final path directly to ensure efficiency. Use the provided MEDIA: line in the chat UI for verification.