MCP Images

You can run an MCP Images server to perform fast image processing, search stock images, and generate AI images. This server operates in standalone mode with a simple npx command, or you can manage it alongside your workspace with MCP Manager for isolated workspaces and visual configuration.

How to use

You use the MCP Images server by running it as an MCP server in either Standalone or Managed mode. In Standalone mode, start the server directly from your project or any workspace and connect your editor or tooling to its HTTP API. In Managed mode, integrate it with MCP Manager to create a separate workspace proxy for each project, enable visual configuration, and securely manage secrets and permissions. The server exposes image processing endpoints, stock image search, and AI image generation capabilities, all coordinated through the MCP SDK for consistent behavior across workspaces.

How to install

# Standalone (via npx)
npx @tscodex/mcp-images@latest

# Global installation (optional)
npm install -g @tscodex/mcp-images

# Managed mode (recommended) requires MCP Manager and its Bridge

Configuration and usage notes

Configure default behavior, providers, image formats, and metadata handling with a configuration file or environment variables as shown below. Secrets are stored securely and accessed via a three-level override system when using MCP Manager.

Configuration file example (standalone behavior): you can place this in your project root as a starting point.

{
  "root": ".",
  "defaultProvider": "pexels",
  "defaultFormat": "webp",
  "defaultMaxWidth": 1920,
  "defaultQuality": 80,
  "saveMetadata": true,
  "embedExif": false
}

Environment variables and secrets

Manage keys and secrets securely. In standalone mode, supply API keys via secret environment variables. In managed mode, use the three-level secret override system to control access.

export SECRET_PEXELS_API_KEY=your_pexels_key
export SECRET_PIXABAY_API_KEY=your_pixabay_key
export SECRET_OPENAI_API_KEY=your_openai_key
export SECRET_OPENAI_ORGANIZATION_ID=your_org_id

Security and permissions

Leverage MCP Manager to enforce OS keychain storage for secrets, process isolation, and granular permissions. You can specify which environment variables, secrets, and AI capabilities a server may access, and you can enable the AI Agent proxy for AI-powered tasks without directly exposing API keys.

AI agent integration

The server supports an AI Agent proxy that lets you generate prompts or interact with AI models without embedding API keys in the server. You can enable AI access per server and monitor token usage from MCP Manager.

Examples

Here are practical usage flows you can follow once the server is running.

Notes

The server is built on the @tscodex/mcp-sdk and relies on the Sharp library for high-performance image processing. It supports image formats like WebP, JPEG, PNG, and AVIF and includes capabilities for resizing, cropping, optimization, watermarking, color extraction, stock image search, and AI image generation.

Available tools

image_process_local

Process local images by resizing, cropping, converting formats, and applying optimizations.

image_analyze

Analyze local images for dimensions, format, file size, and metadata.

image_optimize

Automatically optimize local images, batching compression and format conversion for best results.

image_create_placeholder

Create placeholder images with dimensions displayed.

image_create_favicon

Create favicons in multiple sizes from a source image.

image_add_watermark

Add text or image watermarks to images.

image_apply_filters

Apply visual filters such as blur, sharpen, grayscale, sepia, brightness, and contrast.

image_rotate

Rotate images by a specified angle.

image_crop_custom

Crop images using exact coordinates.

stock_images_search

Search for stock images from providers like Pexels and Pixabay.

stock_images_download_to_project

Download and save stock images to your project.

ai_generate_image

Generate images using AI providers such as OpenAI DALL-E.

image_extract_colors_local

Extract dominant colors and build color palettes from local images.

generate_color_palette_image

Generate a visual color palette image from a local image.