Context-aware image generation

When you need an image in the middle of development or content creation, switching to a separate service breaks your context. The article content, the project background, the brand direction — none of it carries over. You end up re-explaining the same information to a different tool.

With imgx-mcp, the AI you’re already working with uses everything it knows — the article’s purpose, target audience, project direction — to generate the right image. Just say “I need a cover image for this article” and the AI determines the composition, style, and prompt. Review and editing happen in the same session.

Skill — An image production expert in your dev environment

The MCP server gives the AI the ability to generate and edit images. The Skill gives it the knowledge to do it well — so you don’t need to learn prompt syntax, model specifications, or service-specific parameters.

Without the Skill, the AI has powerful tools but limited knowledge of how to use them. With the Skill loaded, the AI draws on a structured body of image production expertise and applies it automatically.

What the Skill brings:

• Automatic prompt construction — You say “I need a cover image.” The AI builds a structured prompt using the Subject-Context-Style framework: identifying what to show, where to place it, and how it should look. No prompt engineering on your part
• 24 editing techniques — Proven approaches for atmosphere, composition, element manipulation, and style transfer. “Make it warmer” or “add depth of field” — the AI selects the right instruction for the model
• Intelligent model selection — Starts with the free model. Suggests paid upgrades only when your needs exceed free tier capabilities — and explains what the upgrade adds. Knows which model has the best text rendering, which supports transparent backgrounds, and which offers the best cost-performance
• Platform-aware sizing — Mention “Twitter OGP” or “App Store screenshot” and the AI selects the correct aspect ratio and resolution. Covers social media, OGP, app stores, print, and blog platforms
• Trending style templates — Ghibli, action figure in box, 3D clay, pixel art, chibi, and more. Say the style name and the AI applies the right prompt structure
• Multi-image consistency — Design tokens and character DNA templates keep slide decks, social media series, and brand assets visually coherent across multiple generations

The image generation models already have these capabilities. The Skill is what makes them accessible without specialized knowledge. It ships with the Plugin and can be installed standalone.

Setup

Add the following to your MCP config file.

Claude Code

.mcp.json in your project root:

{
  "mcpServers": {
    "imgx": {
      "command": "npx",
      "args": ["--package=imgx-mcp", "-y", "imgx-mcp"],
      "env": { "GEMINI_API_KEY": "your-key" }
    }
  }
}

Claude Desktop

claude_desktop_config.json (location):

{
  "mcpServers": {
    "imgx": {
      "command": "npx",
      "args": ["--package=imgx-mcp", "-y", "imgx-mcp"],
      "env": { "GEMINI_API_KEY": "your-key" }
    }
  }
}

Windows: Change to "command": "cmd" and "args": ["/c", "npx", "--package=imgx-mcp", "-y", "imgx-mcp"].

Get your API key from Google AI Studio (free, no credit card required — the default model gemini-2.5-flash-image has a free tier). To use OpenAI, add "OPENAI_API_KEY": "your-key" to env.

For other tools (Gemini CLI, Cursor, Windsurf, etc.), see the usage guide.

Claude Desktop config file location

Restart Claude Desktop after editing.

Supported Tools

Works with any MCP-compatible tool, including:

If your tool supports MCP, the same config works.

MCP Tools

Providers

imgx-mcp itself is free and open source (MIT). The default model (Nano Banana) is free to use — no credit card required. Paid models are available when you need higher resolution or quality. You use your own API keys.

The OGP images and diagrams on this page were themselves produced by imgx-mcp during the content creation session — prompt generation, image creation, iterative editing, and embedding were all handled as a continuous flow.

Links

• GitHub — Source code and issues
• npm — Package
• Usage guide — Setup details and examples
• Story — Why a coffee shop builds image tools