Wyreup for agents.

An MCP server that exposes Wyreup's tools to Claude Code, Cline, Continue, and any MCP-compatible client.

What is MCP

MCP is an open protocol that lets AI assistants call external tools. Wyreup's MCP server wraps every tool so your agent can compress images, redact PDFs, or blur faces on your machine — without your files ever leaving it. Learn more at modelcontextprotocol.io.

Install in your agent runtime

Add the following to your MCP config. Works with Claude Code, Claude Desktop, Cline, Continue, Zed, or any MCP-compatible client. The server runs via npx — no global install required.

MCP server config
{
  "mcpServers": {
    "wyreup": {
      "command": "npx",
      "args": ["-y", "@wyreup/mcp"],
      "env": {
        "WYREUP_API_KEY": "wk_live_..."
      }
    }
  }
}

The WYREUP_API_KEY field is optional — omit it and the free in-browser-class tools still work. Provide it (get one at /account) to unlock the 30 hosted-AI Pro tools (transcription, OCR, translation, text-to-speech, image generation, long-context PDF Q&A, structured-JSON extraction, and more). The agent sees only the tools your config has unlocked, so there's no fail path from missing auth.

  • Claude Desktop (macOS) ~/Library/Application Support/Claude/claude_desktop_config.json
  • Claude Desktop (Windows) %APPDATA%\Claude\claude_desktop_config.json
  • Claude Code (project) .mcp.json in your project root

How agents invoke Wyreup tools

Each Wyreup tool is a named MCP tool. Pass input_paths, output_path (or output_dir for multi-output tools), and params for tool-specific options.

Example MCP tool call
tool: "compress"
arguments:
  input_paths: ["/Users/alice/photo.jpg"]
  output_path: "/Users/alice/photo-compressed.jpg"
  params:
    quality: 80

Pair with the Wyreup skill

The MCP server gives your agent the tools. The skill teaches your AI assistant when and how to use them — recognizing file tasks, choosing the right tool, and handling results correctly.

Install the skill →