Skip to main content
Glama

Gemini Nano Banana MCP

An MCP (Model Context Protocol) server for AI-powered image generation, editing, and video generation using Google Gemini and Veo. Works with Claude Code, Cursor, and any MCP-compatible client.

Features

  • Text-to-Image Generation - Generate images from text prompts via Gemini AI

  • Image Editing - Edit existing images with natural language instructions

  • Reference Images - Use reference images for style and content guidance

  • Text-to-Video Generation - Generate videos from text prompts using Veo (veo-3.1, veo-3, veo-2)

  • Image-to-Video - Use an image as the first frame for video generation

  • First & Last Frame Interpolation - Generate videos between two keyframe images

  • Session Memory - Continue editing the last image without re-specifying the path

  • Configurable Models - Choose any Gemini model for images, any Veo model for videos

  • Media History - Track and browse recently generated images and videos

  • Cross-Platform - Works on macOS, Windows, and Linux

Related MCP server: Gemini Media MCP

Quick Start

1. Get a Gemini API Key

Get your free API key from Google AI Studio.

2. Install

npm install -g @seungmanchoi/nano-banana-mcp

Or install from source:

git clone https://github.com/seungmanchoi/nano-banana-mcp.git
cd nano-banana-mcp
npm install
npm run build

3. Configure Your MCP Client

Claude Code

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "nano-banana": {
      "command": "npx",
      "args": ["-y", "@seungmanchoi/nano-banana-mcp"],
      "env": {
        "GEMINI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Cursor

Add to your MCP settings:

{
  "mcpServers": {
    "nano-banana": {
      "command": "npx",
      "args": ["-y", "@seungmanchoi/nano-banana-mcp"],
      "env": {
        "GEMINI_API_KEY": "your-api-key-here"
      }
    }
  }
}

From Source

If installed from source, use the absolute path:

{
  "mcpServers": {
    "nano-banana": {
      "command": "node",
      "args": ["/absolute/path/to/nano-banana-mcp/dist/index.js"],
      "env": {
        "GEMINI_API_KEY": "your-api-key-here"
      }
    }
  }
}

You can also skip the env field and configure the API key at runtime using the configure_api_key tool.

Authentication Modes

This server supports two authentication modes. The default is API key mode (above).

Mode A — API key (official, default)

Uses a Gemini API key from Google AI Studio. Supports image generation, image editing, and video (Veo). Note that image generation models are largely a paid feature.

Drives your logged-in gemini.google.com session via its session cookies instead of an API key. Free, and supports image generation + editing only (no video).

⚠️ This mode is unofficial. It talks to an undocumented internal endpoint, not the official API. It may break when Google changes things, cookies expire periodically and must be re-extracted, and use is a Terms-of-Service gray area. Intended for personal use with your own account.

1. Extract your cookies from a browser where you're logged into gemini.google.com:

  1. Open https://gemini.google.com and sign in.

  2. DevTools (F12) → Application → Cookies → https://gemini.google.com.

  3. Copy the value of __Secure-1PSID (required) and __Secure-1PSIDTS (recommended).

2a. Configure via environment variables:

{
  "mcpServers": {
    "nano-banana": {
      "command": "npx",
      "args": ["-y", "@seungmanchoi/nano-banana-mcp"],
      "env": {
        "GEMINI_AUTH_MODE": "gemini-web",
        "GEMINI_SECURE_1PSID": "your-__Secure-1PSID-value",
        "GEMINI_SECURE_1PSIDTS": "your-__Secure-1PSIDTS-value"
      }
    }
  }
}

2b. Or configure at runtime with the configure_google_login tool:

Use configure_google_login with secure1psid "<...>" and secure1psidts "<...>"

This switches the active mode to gemini-web and persists to ~/.nano-banana/config.json. Run configure_api_key again at any time to switch back to API key mode.

Model Configuration

Image Models

The default image model is gemini-2.0-flash-preview-image-generation. You can change it in several ways:

Option 1: Environment Variable

Set GEMINI_MODEL in your MCP client config:

{
  "mcpServers": {
    "nano-banana": {
      "command": "npx",
      "args": ["-y", "@seungmanchoi/nano-banana-mcp"],
      "env": {
        "GEMINI_API_KEY": "your-api-key-here",
        "GEMINI_MODEL": "gemini-2.0-flash-preview-image-generation"
      }
    }
  }
}

Option 2: Runtime Tool

Use the configure_model tool to change the default model at runtime. The setting persists across sessions in ~/.nano-banana/config.json.

Set the model to imagen-3.0-generate-002

Option 3: Per-Request Override

Pass the model parameter directly to generate_image, edit_image, or continue_editing to override the default for a single request:

Generate an image of a cat using model imagen-3.0-generate-002

Model Priority

  1. Per-request model parameter (highest priority)

  2. GEMINI_MODEL environment variable

  3. Config file (~/.nano-banana/config.json)

  4. Default: gemini-2.0-flash-preview-image-generation

Available Image Models

Model

Tier

Description

gemini-2.0-flash-preview-image-generation

Free

Default. Native image generation via Gemini 2.0 Flash.

imagen-3.0-generate-002

Paid

Best quality. Google's dedicated image generation model.

imagen-3.0-fast-generate-001

Paid

Fast variant of Imagen 3, optimized for speed.

Note: Free-tier API keys support gemini-2.0-flash-preview-image-generation. Imagen models require billing enabled on your Google Cloud project.

Available Video Models

Model

Description

veo-3.1-generate-preview

Latest. Native audio, scene extension, reference images, 4K support.

veo-3-generate-preview

Previous generation with audio support.

veo-2-generate-preview

Older generation, stable.

For the latest list of models, see Google AI documentation.

Tools

Image Tools

Tool

Description

configure_api_key

Set or update the Gemini API key (switches to apiKey mode). Persists across sessions.

configure_google_login

Switch to free, unofficial gemini-web mode using consumer Gemini cookies (secure1psid, optional secure1psidts). Image generation + editing only.

configure_model

Set the default Gemini model for images (apiKey mode). Persists across sessions.

generate_image

Generate a new image from a text description. Supports optional model override.

edit_image

Edit an existing image with text instructions, optional reference images, and optional model override.

continue_editing

Continue editing the last generated/edited image in the session. Supports optional model override.

list_history

List recently generated and edited images with prompts and timestamps.

Video Tools

Tool

Description

generate_video

Generate a video from a text prompt. Supports text-to-video, image-to-video, and frame interpolation.

list_video_history

List recently generated videos with prompts, models, and timestamps.

Utility Tools

Tool

Description

get_status

Check configuration status, active models, output directories, and last image/video info.

Usage Examples

Image Generation

Generate an image of a sunset over mountains with a lake reflection
Edit the image at ~/nano-banana-images/gen_2025-01-01.png to add a boat on the lake
Continue editing - make the sky more vibrant with orange and pink tones

Video Generation

Generate a video of ocean waves crashing on a rocky shore at sunset
Generate a video of a cat playing with yarn, model veo-3.1-generate-preview, resolution 1080p, duration 8 seconds

Image-to-Video (First Frame)

Generate a video starting from the image at ~/nano-banana-images/gen_2025-01-01.png showing the scene coming to life with wind blowing through the trees

First + Last Frame Interpolation

Generate a video transitioning from the image at ~/images/start.png to ~/images/end.png with a smooth camera pan

Portrait Video

Generate a video of a person walking through a garden, aspect ratio 9:16

History & Status

Show me the last 5 images I generated
Show me recent video history
Check the current status
Switch to imagen-3.0-generate-002 model for higher quality

Video Generation Details

Configuration Options

Parameter

Options

Default

Description

model

veo-3.1-generate-preview, veo-3-generate-preview, veo-2-generate-preview

veo-3.1-generate-preview

Veo model to use

aspectRatio

16:9, 9:16

16:9

Landscape or portrait

resolution

720p, 1080p, 4k

720p

Output resolution

durationSeconds

4, 6, 8 (number)

Varies by model

Video length

numberOfVideos

1+

1

Number of variants

negativePrompt

Any text

-

Elements to avoid

Generation Modes

  1. Text-to-Video: Provide only a prompt

  2. Image-to-Video: Provide prompt + imagePath (used as first frame)

  3. Frame Interpolation: Provide prompt + imagePath (first frame) + lastFramePath (last frame)

Important Notes

  • Video generation takes 1-6 minutes depending on load

  • Generated videos are saved as .mp4 files

  • Videos are watermarked with SynthID technology

  • Pricing: $0.75 per second of generated video

  • Videos are retained on Google servers for 2 days after generation

API Key Configuration

The server loads the API key in the following priority order:

  1. Environment variable - GEMINI_API_KEY

  2. Config file - ~/.nano-banana/config.json

  3. Runtime - via the configure_api_key tool

File Storage

Images

Platform

Path

macOS / Linux

~/nano-banana-images/

Windows

Documents\nano-banana-images\

Videos

Platform

Path

macOS / Linux

~/nano-banana-videos/

Windows

Documents\nano-banana-videos\

Project Structure

src/
├── index.ts              # Entry point
├── server.ts             # MCP server setup and request routing
├── config/
│   └── settings.ts       # API key and model management
├── services/
│   ├── gemini.ts         # Google Gemini & Veo API client
│   └── storage.ts        # Image/video file I/O and history tracking
├── tools/
│   ├── definitions.ts    # MCP tool schemas
│   └── handlers.ts       # Tool request handlers
└── types/
    └── index.ts          # TypeScript type definitions

Development

npm run dev          # Run with tsx (no build needed)
npm run build        # Compile TypeScript
npm run typecheck    # Type check without emitting
npm run lint         # Run ESLint

Tech Stack

  • Runtime: Node.js

  • Language: TypeScript (strict mode, ES2022)

  • MCP SDK: @modelcontextprotocol/sdk

  • AI Models: Google Gemini (images) + Veo (videos)

  • Validation: Zod

License

MIT

Available Tools

10 tools
configure_api_keyA

Set or update the Gemini API key for image generation. The key is stored locally and persists across sessions.

Inputs: apiKey (required).

configure_google_loginA

Switch to the free, unofficial "gemini-web" auth mode that uses your logged-in consumer Gemini (gemini.google.com) session instead of an API key. Supports image generation and editing (no video). Extract the cookies from your browser DevTools (Application > Cookies > gemini.google.com). Cookies are stored locally and may need re-extraction when they expire.

Inputs: secure1psid (required), secure1psidts.

configure_modelA

Set the default Gemini model for image generation and editing. Persists across sessions. Use "quality" param to set either "high" (default) or "fast" model tier. High-quality: gemini-3.1-flash-image-preview (default). Fast/cheap: gemini-2.5-flash-image (default fast).

Inputs: model (required), quality.

continue_editingA

Continue editing the last generated or edited image. Automatically uses the most recent image from the session.

Inputs: prompt (required), referenceImages, model, quality.

edit_imageB

Edit an existing image based on text instructions. Provide the file path of the image to modify.

Inputs: imagePath (required), prompt (required), referenceImages, model, +1 more.

generate_imageA

Generate a new image from a text description using Gemini AI. Returns the generated image and saves it to disk.

Inputs: prompt (required), model, quality.

generate_videoA

Generate a video from a text prompt using Gemini Veo. Supports text-to-video, image-to-video (first frame), and first+last frame interpolation. Video generation takes 1-6 minutes. Available models: veo-3.1-generate-preview (latest), veo-3-generate-preview, veo-2-generate-preview.

Inputs: prompt (required), model, imagePath, lastFramePath, +5 more.

get_statusA

Check the current configuration status, active model, and last image/video information.

Inputs: no named parameters.

list_historyA

List recently generated and edited images with their prompts and timestamps.

Inputs: count.

list_video_historyA

List recently generated videos with their prompts, models, and timestamps.

Inputs: count.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4/5.0

    Scored across 10 tools

    Disambiguation5/5

    Each tool targets a distinct operation: configuration tools are separated by purpose (API key, Google login, model), image tools separate generation from editing and continuation, video tool stands alone, and info tools provide status and history without overlap.

    Naming Consistency5/5

    All tools follow a consistent verb_noun snake_case pattern (e.g., configure_api_key, generate_image, list_history). The only slight deviation is 'continue_editing' (gerund) but it still fits the verb+noun structure.

    Tool Count5/5

    The server covers configuration (3 tools), image generation/editing (3), video generation (1), and info retrieval (3) with a total of 10 tools. This is well-scoped for a Gemini AI media generation server without unnecessary bloat.

    Completeness4/5

    The tool surface covers the full lifecycle: configuration, generation, editing, and history. A minor gap is the lack of delete or management actions for generated media, but the core workflow is complete.

    Maintenance

    ActivityNo data
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers