I built an MCP server that lets Claude (Desktop or Code) watermark images using natural language
prompts.
Setup takes 10 seconds - add this to your Claude config:
{"mcpServers":{"markly":{"command":"npx","args":["-y","markly-mcp-server"]}}}
Then ask Claude things like:
- "Add 'Copyright 2026' to this image, bottom-right, semi-transparent"
- "Watermark all images in /photos with my logo.png at 20% opacity"
- "Put a tiled DRAFT watermark across this proof sheet"
How it works: The MCP server exposes 4 tools (text watermark, logo watermark, batch, usage check)
that call the Markly REST API. Images are processed server-side with Imagick and returned
immediately. Nothing is stored after processing.
Pricing: Free tier needs no API key, no signup. Output has a small "markly.cloud" mention. Credit
packs start at 5 EUR for 250 images with no branding.
The MCP server is MIT licensed. The REST API is also available directly if you prefer to integrate
programmatically.
Links:
- npm: https://www.npmjs.com/package/markly-mcp-server
- API docs: https://www.markly.cloud/developers
- Website: https://www.markly.cloud
Built with Laravel (API), Imagick (image processing), TypeScript (MCP server). Feedback welcome.
I built an MCP server that lets Claude (Desktop or Code) watermark images using natural language prompts.