MCP Servers

A collection of Model Context Protocol servers, templates, tools and more.

N
Nanobanana MCP Server
by @ryaker
GitHub Repo(1 stars)

MCP server for AI image generation with Gemini — character consistency, object fidelity, grounding, and thinking mode

Created 3/9/2026
Updated about 10 hours ago
README
View Source
Repository documentation and setup instructions

Nano Banana MCP Server

Nano Banana MCP Server 🍌

A production-ready Model Context Protocol (MCP) server for AI-powered image generation using Google's Gemini image models — fast iteration, high-resolution output, character consistency, Google Search grounding, and thinking mode.

✨ Features

Three-Tier Model Support

| Model | ID | Speed | 1K / 2K / 4K | Best For | |-------|----|-------|--------------|----------| | 🍌 Nano Banana 2 | gemini-3.1-flash-image-preview | 4–6s | $0.067 / $0.101 / $0.151 | Default — object fidelity, grounding, thinking, 14 ARs | | 💎 Nano Banana Pro | gemini-3-pro-image-preview | 10–20s | $0.067 / $0.134 / $0.240 | Character consistency, max lighting/texture fidelity, 4K | | ⚡ Nano Banana | gemini-2.5-flash-image | 10–15s | $0.039 flat | Budget drafts, high-volume workflows |

0.5K draft tier available on Nano Banana 2 at $0.045/image. 50% batch discount available on all models.

Nano Banana 2 — Default Model (gemini-3.1-flash-image-preview)

The right choice for most tasks. 3–5× faster than Pro, exclusive features:

  • 🎁 Object Fidelity — reproduce products, logos, and brand assets with precision across scenes; up to 10 object reference images
  • 🖼️ Photo Editing with Identity Preservation — change background, outfit, or setting while keeping a person's exact features
  • 🔍 Image Search Grounding — retrieves real-world visual references from Google during generation; improves accuracy for landmarks, real people, and brand logos (NB2-exclusive)
  • 🧠 Thinking Mode — thinking_level="low" or "high" for complex layouts, text-heavy images, and multi-element compositions (NB2-exclusive)
  • 📝 Text in Images — strong short-text rendering; use JSON-structured prompts for precise infographic copy and labels
  • 🌍 Multilingual Text — generate and translate text within images across 8+ languages
  • 📐 14 Aspect Ratios — including extreme formats: 4:1 banners, 1:4 strips, 8:1 ultra-wide, 1:8 ultra-tall
  • 📏 0.5K → 4K Resolution — four output tiers from fast drafts to print quality

Nano Banana Pro — When to Use It

Pro's strength is character consistency across people and characters plus max lighting/texture fidelity:

  • 👤 Character Consistency — up to 5 characters maintained across a series with higher fidelity; preferred for protagonist-driven storyboards and campaign mascots
  • 💡 Complex lighting and materials — intricate shadow relationships, reflective surfaces, photorealistic skin
  • 🖨️ 4K print-quality output — highest resolution for physical production
  • 14 total references (5 character + 10 object) combined

Core Features

  • 🎨 Generate or Edit — create from scratch or modify existing images with natural language
  • 📁 Named References — configure brand assets once (REFERENCE_REGISTRY), reference by name forever
  • 🛡️ Production Ready — asset deduplication, error handling, structured logging
  • ⚡ MCP Native — works with Claude.ai, Claude iOS, Claude Desktop, Claude Code, Cursor, and any MCP client

🚀 Quick Start

📚 New to nano banana? Check out the Complete Usage Guide with detailed examples and best practices!

⚡ Need quick reference? See the Quick Reference Card for common patterns.

📋 Want code examples? Browse Practical Examples for copy-paste ready code.

Prerequisites

  1. Google Gemini API Key - Get one free here
  2. Python 3.11+ (for development only)

Installation

Option 1: From MCP Registry (Recommended) This server is available in the Model Context Protocol Registry. Install it using your MCP client.

mcp-name: io.github.zhongweili/nanobanana-mcp-server

Option 2: Using uvx

uvx nanobanana-mcp-server@latest

Option 3: Using pip

pip install nanobanana-mcp-server

🔧 Configuration

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "nanobanana": {
      "command": "uvx",
      "args": ["nanobanana-mcp-server@latest"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
    }
  }
}

Configuration file locations:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Claude Code (VS Code Extension)

Install and configure in VS Code:

  1. Install the Claude Code extension
  2. Open Command Palette (Cmd/Ctrl + Shift + P)
  3. Run "Claude Code: Add MCP Server"
  4. Configure: 
    {
  "name": "nanobanana",
  "command": "uvx",
  "args": ["nanobanana-mcp-server@latest"],
  "env": {
    "GEMINI_API_KEY": "your-gemini-api-key-here"
  }
}

Cursor

Add to Cursor's MCP configuration:

{
  "mcpServers": {
    "nanobanana": {
      "command": "uvx",
      "args": ["nanobanana-mcp-server@latest"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
    }
  }
}

Continue.dev (VS Code/JetBrains)

Add to your config.json:

{
  "mcpServers": [
    {
      "name": "nanobanana",
      "command": "uvx",
      "args": ["nanobanana-mcp-server@latest"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
    }
  ]
}

Open WebUI

Configure in Open WebUI settings:

{
  "mcp_servers": {
    "nanobanana": {
      "command": ["uvx", "nanobanana-mcp-server@latest"],
      "env": {
        "GEMINI_API_KEY": "your-gemini-api-key-here"
      }
    }
  }
}

Gemini CLI / Generic MCP Client

# Set environment variable
export GEMINI_API_KEY="your-gemini-api-key-here"

# Run server in stdio mode
uvx nanobanana-mcp-server@latest

# Or with pip installation
python -m nanobanana_mcp_server.server

🌐 Self-Hosting with Cloudflare Tunnel (Claude.ai + Claude iOS)

Want to run Nano Banana on your home computer and access it from Claude.ai (web) and Claude iOS — from anywhere?

Use a free Cloudflare Tunnel — no port forwarding, no static IP, no VPS required.

Claude.ai / Claude iOS
        │  HTTPS
        ▼
Cloudflare Tunnel (free)
        │  localhost:9000
        ▼
Your Home Computer

Quick setup:

# 1. Start server in HTTP mode
FASTMCP_TRANSPORT=http uvicorn asgi:app --host 0.0.0.0 --port 9000

# 2. Install cloudflared and create tunnel
brew install cloudflare/cloudflare/cloudflared   # macOS
cloudflared tunnel login
cloudflared tunnel create nanobanana
cloudflared tunnel route dns nanobanana mcp.yourdomain.com
cloudflared tunnel run nanobanana

# 3. Add to Claude.ai: Settings → Integrations → Add Integration
#    URL: https://mcp.yourdomain.com/mcp

Full step-by-step guide: docs/CLOUDFLARE_TUNNEL.md

Covers: macOS launchd, Linux systemd, Windows NSSM, Cloudflare Access auth, and troubleshooting.

🛠️ Claude Code Skills

If you use Claude Code, install the included workflow skills to get smarter, context-aware image generation — no need to explain the server every session.

# Install all skills
for skill in image-generation-workflow ai-image-model-selector image-reference-workflow; do
  mkdir -p ~/.claude/skills/$skill
  cp skills/${skill}.md ~/.claude/skills/$skill/SKILL.md
done

| Skill | Triggers | |-------|---------| | image-generation-workflow | "generate image", "create picture" | | ai-image-model-selector | "which model", "best for character consistency" | | image-reference-workflow | "use this photo", "reference image" |

See skills/README.md for customization instructions.

⚙️ Environment Variables

Optional configuration:

# Required
GEMINI_API_KEY=your-gemini-api-key-here

# Optional
IMAGE_OUTPUT_DIR=/path/to/image/directory  # Default: ~/nanobanana-images
FASTMCP_TRANSPORT=http                     # "stdio" (local) or "http" (remote/tunnel)
FASTMCP_HOST=0.0.0.0                      # Bind address for HTTP mode
FASTMCP_PORT=9000                          # Port for HTTP mode
LOG_LEVEL=INFO                             # DEBUG, INFO, WARNING, ERROR
LOG_FORMAT=standard                        # standard, json, detailed

See .env.example for all options with descriptions.

🐛 Troubleshooting

Common Issues

"GEMINI_API_KEY not set"

  • Add your API key to the MCP server configuration in your client
  • Get a free API key at Google AI Studio

"Server failed to start"

  • Ensure you're using the latest version: uvx nanobanana-mcp-server@latest
  • Check that your client supports MCP (Claude Desktop 0.10.0+)

"Permission denied" errors

  • The server creates images in ~/nanobanana-images by default
  • Ensure write permissions to your home directory

Development Setup

For local development:

# Clone repository
git clone https://github.com/ryaker/nanobanana-mcp-server.git
cd nanobanana-mcp-server

# Install with uv
uv sync

# Set environment
export GEMINI_API_KEY=your-api-key-here

# Run locally
uv run python -m nanobanana_mcp_server.server

📄 License

MIT License - see LICENSE for details.

🆘 Support

Quick Setup
Installation guide for this server

Install Package (if required)

uvx nanobanana-mcp-server

Cursor configuration (mcp.json)

{ "mcpServers": { "ryaker-nanobanana-mcp-server": { "command": "uvx", "args": [ "nanobanana-mcp-server" ] } } }
Author Servers
Other servers by ryaker

MCP Mem0 General

OpenAI just added memory across your chats across your openAI account. But wouldn't it be awesome to have general AI memory across all your interactions with any and all AI tools, IDEs, chatbots.... Now if it supports MCP you can with https://mem0.ai/ Give Claude desktop memory. Give cursor or windsurf memory across sessions or different projects.

8

MCP Js Server Template

Simplified MCP Server Template for Node.js

2

Outlook MCP

MCP server for Claude to access Outlook data via Microsoft Graph API

62