---
title: "mdfy MCP server: setup in 4 lines"
url: https://mdfy.app/PrxR1Uj4
updated: 2026-05-14T18:15:49.480Z
source: "mdfy.app"
---
# mdfy MCP server: setup in 4 lines

> For Claude Desktop, Cursor, and any other MCP-enabled tool.

## What MCP gives us

MCP (Model Context Protocol) is Anthropic's open standard for letting AI tools call external services. The mdfy MCP server exposes seven tools — read, search, create, update, publish, delete, list — that any MCP host can call. The result: the AI doesn't just *read* your hub URL; it can actively manage what's in it.

## Setup

Add this to your MCP config (Claude Desktop: `~/Library/Application Support/Claude/mcp_config.json`; Cursor: `.cursor/mcp.json` in the project):

```json
{
  "mcpServers": {
    "mdfy": {
      "command": "npx",
      "args": ["-y", "@mdfy/mcp-server"]
    }
  }
}
```

Restart the host (Claude Desktop or Cursor). The mdfy tools show up in the tool palette.

## Authentication

The MCP server reads `MDFY_EDIT_TOKEN` and `MDFY_USER_EMAIL` from the environment. Add them to the MCP config like this:

```json
{
  "mcpServers": {
    "mdfy": {
      "command": "npx",
      "args": ["-y", "@mdfy/mcp-server"],
      "env": {
        "MDFY_USER_EMAIL": "you@example.com",
        "MDFY_EDIT_TOKEN": "..."
      }
    }
  }
}
```

You can copy the edit token from the Share modal of any doc you own; it scopes to your whole account, not a single doc.

## The tools exposed

| Tool | Purpose |
|---|---|
| `mdfy_read` | Fetch a doc / bundle / hub by id or URL |
| `mdfy_search` | Hybrid recall across your hub |
| `mdfy_create` | Create a new doc with a body |
| `mdfy_update` | Patch an existing doc (with optional change summary) |
| `mdfy_publish` | Flip a draft doc to public |
| `mdfy_delete` | Soft-delete (recoverable from trash) |
| `mdfy_list` | List all your docs/bundles, optionally filtered |

## What changes in your workflow

Before MCP: you ask Claude "summarise the launch plan", and Claude says "I'd need to read the plan — paste it for me." After MCP: Claude calls `mdfy_read` with the doc id, gets the markdown, and answers.

That's the deal. The tool palette is small (7 tools) but the reach is big — every operation you'd do in the editor is also callable from the AI.