We just shipped something we’ve been quietly excited about — a Model Context Protocol (MCP) server that lets you use maskr.io directly inside AI assistants like Claude, Cursor, Windsurf, and any other MCP-compatible client.
No browser tabs. No copy-pasting URLs. No switching between tools. Just ask your AI assistant to remove a background or upscale an image, and it happens right there in the conversation.
What Is MCP?
Model Context Protocol is an open standard that lets AI assistants connect to external services. Think of it as a way for AI models to use tools — not by pretending, but by actually calling real APIs on your behalf.
Anthropic introduced MCP as a universal standard for AI-to-service communication, and it’s been adopted across the industry — from Claude to Cursor to Windsurf and beyond. We built a server that speaks this protocol, so any MCP-compatible client can tap into maskr.io natively.
What You Can Do
Once connected, your AI assistant gets access to the following tools:
Remove backgrounds — send an image URL, get back a transparent result
Upscale images — 2x or 4x, with output in PNG, JPEG, or WebP
Check your balance — see how many credits you have left
Check pricing — find out what a specific operation will cost before running it
Browse plans and packages — compare subscription tiers or credit packs
It uses your existing maskr.io account for authentication. Same credits, same account, same pricing — just a new way to access it.
How It Works
You don’t need to understand the internals to use it, but here’s the short version for the curious.
The MCP server sits between the AI assistant and our existing API. When you ask it to remove a background, here’s what happens behind the scenes:
The assistant sends a tool call to our MCP server
The server validates your identity via OAuth
It downloads the image, sends it to our processing pipeline
The GPU does its thing (background removal, upscaling — whatever you asked for)
The result comes back as an inline image with a download link
The whole thing runs on Streamable HTTP — stateless, fast, and horizontally scalable. No WebSocket sessions to manage, no state to lose. Each request is independent.
Good to Know
There are a couple of limitations worth mentioning — not because we cut corners, but because of how the MCP protocol works today.
You need to provide a public image URL. You can’t drag an image into the chat and ask the assistant to process it. MCP tools communicate through structured text — there’s currently no mechanism in the protocol for passing files or binary data from the conversation to an external tool. So instead of uploading an image directly, you give the assistant a link (e.g., from your CDN, a cloud storage bucket, or any publicly accessible URL) and it takes it from there.
Results depend on the client. Most MCP clients will display the processed image inline and provide a download link. But the exact experience — how the image is shown, whether you can click to download — varies depending on which client you’re using. We include a direct download URL with every result, so you’ll always have access to the full-resolution file regardless.
We’re keeping an eye on the MCP spec as it evolves. As the protocol adds support for richer input types, we’ll be ready to support them.
Getting Started
Claude.ai
Open Settings → Integrations
Search for Maskr in the integration directory
Click Connect — you’ll be redirected to sign in with your maskr.io account
Authorize the connection, and you’re done
No API keys to copy, no tokens to manage. The OAuth flow is handled automatically — you just log in once and it works.
Claude Desktop, Cursor, Windsurf, and Other MCP Clients
Add this to your MCP configuration:
{
"mcpServers": {
"maskr": {
"type": "streamable-http",
"url": "https://mcp.maskr.io/mcp"
}
}
}
Your client will automatically discover the auth requirements and walk you through signing in when you first use a maskr.io tool.
What You Can Ask
Once connected, just talk naturally:
"Remove the background from this image: https://example.com/photo.jpg"
“Upscale this image 4x and give me a PNG”
“How many credits do I have left?”
“How much would it cost to remove the background from a 2000x1500 image?”
“What subscription plans do you offer?”
Your assistant handles the rest.
Pricing
Same as everything else on maskr.io — it uses your existing credits. No MCP surcharge, no separate billing. Background removal and upscaling cost the same whether you use the web app, the API, or an AI assistant.
Check the pricing page for the full breakdown.
What’s Next
This is our first MCP release, and we’re already working on bringing more services to it — object removal, compression, and batch processing are on the roadmap.
Got questions or ideas? Drop by the community or find us on X.
Go process some images. 🚀