# mcpify MCP server

> Converts any OpenAPI REST API into a production-ready MCP server for use with AI clients like Claude Code and Cursor.

- Repository: https://github.com/furkan708/mcpify
- Page: https://hostedmcp.org/servers/furkan708/mcpify
- Transport: stdio
- Stars: 2
- License: MIT
- Latest version: 1.19.0

## Install

### Claude Code

```sh
claude mcp add mcpify -- uvx mcpify-openapi
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "mcpify": {
      "command": "uvx",
      "args": [
        "mcpify-openapi"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "mcpify": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "mcpify-openapi"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"mcpify","type":"stdio","command":"uvx","args":["mcpify-openapi"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "mcpify": {
      "command": "uvx",
      "args": [
        "mcpify-openapi"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add mcpify -- uvx mcpify-openapi
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.mcpify]
command = "uvx"
args = ["mcpify-openapi"]
```

### Gemini CLI

```sh
gemini mcp add -s user mcpify uvx -- mcpify-openapi
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "mcpify": {
      "command": "uvx",
      "args": [
        "mcpify-openapi"
      ]
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `OAUTH2_CLIENT_ID` | no | no |  |
| `OAUTH2_CLIENT_SECRET` | no | no |  |
| `PETSTORE_KEY` | no | no |  |

## FAQ

### How do I turn an OpenAPI spec into an MCP server?

You can use the command `mcpify serve <url-to-spec>` to immediately start serving the API as MCP tools.

### Can I use this with existing authentication like OAuth2?

Yes, it has built-in support for OAuth2 client-credentials, as well as bearer, basic, header, and query authentication styles.

### How can I prevent the AI from seeing sensitive information?

You can use the `--redact` flag to mask specific keys like passwords or tokens in API responses so the model never sees them.

### Does it support multiple APIs at once?

Yes, you can define several APIs in a single `.mcpify.toml` configuration file to serve them as a single tool surface.