Nano Banana Pro

After installing, open the MyClaw/OpenClaw workspace where you want the asset. Ask the agent to use Nano Banana Pro for a specific job, such as generate a 1K blog hero for this article, create a product mockup, or edit this image and remove the background object.

Review the generated PNG inside the workspace, ask the agent for targeted changes, and only switch to 2K or 4K when the prompt is locked. Then have MyClaw place the final image into your app, post, documentation, or automation output.

• You want a nano banana pro skill Claude Code or Codex can call during a coding, content, or design workflow.
• You need a nano banana pro agent skill that turns prompts, local references, or uploaded images into usable visual assets.
• You want to generate drafts quickly at 1K, refine the prompt or edit request, then export a 2K or 4K final.
• You are running an OpenClaw agent on MyClaw and want image generation available in the same environment as your files, app, blog, or automation task.

• You need a different image model such as DALL-E, Stable Diffusion, or Flux instead of Google's Gemini image model.
• You do not have a valid GEMINI_API_KEY or image generation access for your Google AI account.
• You need manual layer-by-layer design editing, typography tuning, or vector illustration controls.
• You need guaranteed exact text rendering in complex images without reviewing and iterating on the output.

Install the same image workflow into Claude Code-style, Codex, or OpenClaw agent environments. The agent can generate images while it builds a page, prepares a blog post, designs assets, or updates a project in the same workspace.

Use Nano Banana Pro directly from a hosted MyClaw agent instead of switching between a chat tool, an image app, and your project files. The generated PNG is saved where the agent can inspect it, rename it, reuse it, or attach it to the next task.

Describe the subject, composition, lighting, style, background, and output size, then let the skill create a fresh image. This works well for hero images, product concepts, editorial illustrations, thumbnail ideas, and visual prototypes.

Pass an existing image and describe what should change, such as removing an object, adjusting colors, changing the style, or improving a product mockup. The prompt template is designed to preserve everything else unless you ask for a broader transformation.

Start with fast 1K drafts, move to 2K when the composition is close, and use 4K for final assets. The skill maps common phrases such as high-res or ultra quality to the correct model setting.

Outputs are named with a timestamp and prompt-derived filename, which keeps agent runs organized. That makes it easier to compare drafts, choose a final asset, and reference the exact image in a follow-up coding or publishing step.

Ask your MyClaw agent to create a blog cover, landing page visual, product mockup, or social preview image, then place the generated asset into the project it is already editing.

Use the nano banana pro skill Claude Code or Codex can invoke when a coding task needs real visual output, such as app screenshots, empty-state artwork, marketing graphics, or design references.

Provide an input image and ask the agent to make a focused change: clean up a background, change a style, remove a distracting object, recolor an asset, or create a polished variant.

Add image generation to repeatable AI agent workflows such as content publishing, product documentation, campaign production, ecommerce mockups, or design QA where each run needs a fresh visual asset.