Resize & crop
Scale by width, height, percentage, or fit mode. Crop by coordinates, dimensions, or gravity with a single tool call.
Open-source image manipulation MCP server
Give your agents a complete image toolkit
img-mcp is an MCP server that lets agents resize, crop, convert, compress, and edit images locally with Node.js — plus generate images when you want AI. No Docker, no shared keys, no friction.
Works with
Claude Desktop
Cursor
Windsurf
Kimi Code
OpenCode
Codex CLI
Features
A complete image toolkit for agents
From fast local edits to AI-powered generation — one MCP server, zero setup.
Convert & compress
Switch between JPEG, PNG, WebP, AVIF, GIF, and more. Optimize quality and file size for any channel.
Batch operations
Apply the same transform to multiple images in one call. Perfect for thumbnails, exports, and asset pipelines.
Metadata & EXIF
Read, write, and strip image metadata. Keep or remove EXIF, IPTC, and XMP data on demand.
Text-to-image
Generate high-quality images from natural language prompts with full control over aspect ratio and resolution.
Reference images & transparency
Guide AI generation with base64 reference images and produce clean transparent assets for overlays.
How it works
From command to edited image in three steps
img-mcp translates your agent's instructions into local image operations or generation calls.
1
Set up in seconds
No API key is needed for local manipulation. Add a Google API key only if you want AI image generation.
2
Add to Claude Code
Run one command to register img-mcp as an MCP server in Claude Code. It also works in Cursor, Windsurf, Kimi, and other MCP hosts.
3
Edit or generate images
Tell your agent what to do. img-mcp resizes, converts, crops, or generates images and returns the result.
Quick start
Add to Claude Code in one command
No JSON files, no Docker, no manual config. Just run the CLI.
terminal
# Local image manipulation only (no API key needed) $ claude mcp add --transport stdio img-mcp -- npx -y img-mcp # With AI image generation enabled $ claude mcp add --env GOOGLE_API_KEY=YOUR_KEY --transport stdio img-mcp -- npx -y img-mcp
Roadmap
Where img-mcp is headed
AI generation brought us here. The future is fast, local, non-AI image manipulation built on Node.js.
Text-to-image generation
Generate images from prompts with size, ratio, and output controls.
Reference-image generation
Use existing images to guide style, subject, and composition.
Transparent backgrounds
Automatic background removal for clean asset generation.
Resize & convert
Scale images and switch formats locally with no API call.
Crop, rotate & flip
Coordinate-based cropping, rotation, mirroring, and EXIF auto-orient.
Compress & optimize
Quality tuning and target file-size optimization for any format.
Batch operations
Apply transforms to multiple images in a single tool call.
Metadata & EXIF
Read, write, and strip image metadata on demand.
Watermark & overlay
Add text or image watermarks with positioning and opacity.
Upscale & super-resolution
Increase resolution and recover detail when quality matters.
Inpainting & outpainting
Edit parts of an image or extend the canvas beyond its edges.
Additional providers
Support for more image models beyond Google Nano Banana.
Ready to give your agent image superpowers?
Join the open-source community and start manipulating and generating images with img-mcp today.