Peelaway MCP server

Peelaway is a hosted MCP image-editing server for AI agents. It exposes one production workflow for removing objects, removing backgrounds, and making guided generative edits, with asynchronous status and result retrieval.

Open the setup guide

Connect

Use the hosted Streamable HTTP endpoint—there is no package or local server to install:

https://mcp.peelaway.io/mcp

MCP clients that support OAuth can discover Peelaway's OAuth 2.1 metadata from the endpoint, open the browser sign-in flow, and retain the resulting token. Clients without OAuth support can send a Peelaway API key as a bearer token.

Claude Code

claude mcp add --transport http peelaway https://mcp.peelaway.io/mcp

Then open /mcp in a Claude Code session and authenticate when prompted.

Cursor and other HTTP-native clients

Copy peelaway.mcp.json into the MCP configuration format your client expects:

{
  "mcpServers": {
    "peelaway": {
      "url": "https://mcp.peelaway.io/mcp"
    }
  }
}

For API-key authentication, add an authorization header:

{
  "mcpServers": {
    "peelaway": {
      "url": "https://mcp.peelaway.io/mcp",
      "headers": {
        "Authorization": "Bearer pk_..."
      }
    }
  }
}

Never commit a real pk_ key. Prefer OAuth when the client supports it.

Tools

Tool	Purpose
edit_image	Submit an image and editing instruction; returns a job ID.
get_job_status	Read the job's pending, done, or error state.
get_job_result	Retrieve the finished image's signed result URL.

The editing instruction selects object removal, background removal, or generative editing behind the same tool contract. JPEG, PNG, and WebP inputs are supported. Background removal returns PNG so transparency is preserved.

Authentication and credits

Create an account and verify its email before the first edit. New verified accounts receive 10 free credits; each edit spends one credit and a failed job is refunded automatically. API keys are created from the Peelaway dashboard.

Connect Peelaway and review client-specific instructions

Machine-readable discovery

• Official MCP Registry entry
• MCP manifest
• OAuth protected-resource metadata
• OpenAPI specification
• Agent instructions
• LLM-readable overview

Repository scope

This repository contains public connection documentation and configuration examples for the hosted Peelaway service. It intentionally does not contain the private service source code, credentials, or deployment configuration.

The MIT license applies only to the documentation and configuration in this repository. It does not license the hosted Peelaway service or its private source code. Using the service remains subject to Peelaway's published terms and privacy policy.

Documentation problems and reproducible connection failures are welcome as GitHub issues. Report security vulnerabilities through GitHub's private vulnerability reporting instead of a public issue.