Nano Banana

mcp for nano banana using openrouter.ai

typescript

5

GitHub Stars

typescript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "aeven-ai-mcp-nanobanana": {
      "command": "npx",
      "args": [
        "-y",
        "@aeven/nanobanana-mcp@latest"
      ],
      "env": {
        "MODEL_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Nano Banana MCP Server is a specialized extension that enables you to generate, edit, restore, and manage images through any MCP-compatible client. It connects to a model hosted by providers such as OpenRouter by default and exposes a streamlined set of image tools you can trigger directly from your MCP client for fast, automated visual results.

How to use

You interact with Nano Banana through your MCP client by invoking image-generation commands. Use the provided commands to create new images from descriptive prompts, edit existing images with natural language instructions, restore damaged photos, and generate icons, patterns, stories, diagrams, and more. The server handles communication with the image model and returns 1024×1024 images that you can save and organize with smart filenames.

Key actions include generating a single image, generating multiple variations, editing a photo, restoring an image, and producing non-image assets like icons and diagrams. Your MCP client will expose endpoints such as generate_image, edit_image, restore_image, icon, pattern, story, diagram, and nanobanana for the natural-language interface.

How to install

Prerequisites: you need an MCP-compatible CLI, Node.js 18+, npm, and an API key from a model provider (for example, OpenRouter) to access google/gemini-2.5-flash-image.

Install via your MCP client’s extension manager or by registering the MCP server with your CLI. The following are representative installation flows you can follow depending on your toolchain.

For Codex CLI use the following command to install and register the MCP server, providing your API key at setup time.

Additional setup and configuration

Environment variables are used to authenticate with your model provider. The required variable shown here is MODEL_API_KEY. You set this either during installation or in your shell environment, depending on your client. The server expects this key to access the cloud provider hosting the model.

If you are using a provider other than the default, you can override the model endpoint by setting MODEL_BASE_URL, MODEL_ID, MODEL_REFERER, MODEL_TITLE, and MODEL_GENERATE_PATH as needed.

Troubleshooting

If you encounter command recognition issues, ensure the MCP server is properly registered in your client and that you have restarted the client after configuration changes.

Make sure your API key is correctly configured as MODEL_API_KEY in your environment or via your client’s setup flow. If an image cannot be found, verify input files exist in the searched directories and that you are using valid input paths.

Available tools

generate_image

Create a new image from a text prompt or a natural-language instruction using the connected image model.

edit_image

Modify an existing image with a natural-language instruction, returning the edited result.

restore_image

Restore or enhance an old or damaged image by applying restoration techniques via the model.

icon

Generate icons, favicons, and UI assets in multiple sizes and formats.

pattern

Create seamless patterns or textures suitable for backgrounds and design elements.

story

Produce a sequence of images that tell a visual story or demonstrate a process.

diagram

Generate technical diagrams, flowcharts, and architectural mockups from text prompts.

nanobanana

Provide a natural-language interface to generate and manipulate images through MCP commands.