> ## Documentation Index
> Fetch the complete documentation index at: https://docs.make87.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Model Context Protocol Integration

> Integrate m87 with AI agents using the Model Context Protocol (MCP)

The m87 CLI includes a built-in [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server, enabling AI agents to interact with your edge devices and use all m87 platform commands as tools.

## Overview

The MCP integration exposes m87's functionality to AI agents, allowing them to:

* Query device status and metrics
* Execute commands on remote devices
* Manage deployments
* Forward ports
* Access logs and audit trails
* Transfer files
* And more

This enables natural language device management through AI assistants like Claude.

## Setup

<Steps>
  <Step title="Verify m87 Installation">
    Ensure m87 is installed and accessible in your PATH:

    ```bash theme={null}
    which m87
    ```

    If not installed, follow the [installation guide](/installation).
  </Step>

  <Step title="Configure Your MCP Client">
    Add m87 to your MCP client configuration. The configuration varies by client:

    ### Claude Desktop / Claude Code

    Edit your MCP settings file:

    **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`

    **Linux**: `~/.config/Claude/claude_desktop_config.json`

    Add the following configuration:

    ```json theme={null}
    {
      "mcpServers": {
        "m87": {
          "type": "stdio",
          "command": "m87",
          "args": ["mcp"]
        }
      }
    }
    ```
  </Step>

  <Step title="Authenticate">
    Ensure you're logged in to m87:

    ```bash theme={null}
    m87 login
    ```

    The MCP server will use your existing authentication.
  </Step>

  <Step title="Restart Your MCP Client">
    Restart Claude Desktop or your MCP client to load the new configuration.
  </Step>
</Steps>

## Configuration Options

### Using Absolute Path

If `m87` is not in your PATH, specify the full path to the binary:

```json theme={null}
{
  "mcpServers": {
    "m87": {
      "type": "stdio",
      "command": "/usr/local/bin/m87",
      "args": ["mcp"]
    }
  }
}
```

To find the full path:

```bash theme={null}
which m87
```

### Custom Environment Variables

You can pass environment variables to the MCP server:

```json theme={null}
{
  "mcpServers": {
    "m87": {
      "type": "stdio",
      "command": "m87",
      "args": ["mcp"],
      "env": {
        "RUST_LOG": "debug"
      }
    }
  }
}
```

## Usage Examples

Once configured, you can interact with your devices through natural language in your AI assistant:

### Device Management

<CodeGroup>
  ```text Example Prompt theme={null}
  "List all my connected devices"
  ```

  ```text Example Prompt theme={null}
  "Show me the status of my production-server"
  ```

  ```text Example Prompt theme={null}
  "Check the health and any incidents on device-123"
  ```
</CodeGroup>

### Remote Execution

<CodeGroup>
  ```text Example Prompt theme={null}
  "Run 'systemctl status nginx' on production-server"
  ```

  ```text Example Prompt theme={null}
  "Execute 'df -h' on all devices and summarize the disk usage"
  ```
</CodeGroup>

### Logs and Monitoring

<CodeGroup>
  ```text Example Prompt theme={null}
  "Show me the latest logs from my edge-device-01"
  ```

  ```text Example Prompt theme={null}
  "Get system metrics for production-server"
  ```

  ```text Example Prompt theme={null}
  "Check who accessed my device in the last 24 hours"
  ```
</CodeGroup>

### Deployments

<CodeGroup>
  ```text Example Prompt theme={null}
  "Show me the deployment status for production-server"
  ```

  ```text Example Prompt theme={null}
  "What containers are running on edge-device-01?"
  ```
</CodeGroup>

## Available MCP Tools

The m87 MCP server exposes all platform commands as tools, including:

* **Device Management**: List devices, check status, approve devices
* **Remote Access**: Execute commands, open shells, forward ports
* **File Operations**: Copy files, sync directories
* **Container Management**: Docker passthrough, view containers
* **Monitoring**: View logs, metrics, audit trails
* **Deployments**: Deploy compose files, check deployment status
* **Serial Access**: Forward serial mounts

The AI agent will automatically select and use the appropriate tools based on your requests.

## Supported MCP Clients

The m87 MCP server works with any MCP-compatible client:

* [Claude Desktop](https://claude.ai/download)
* [Claude Code (VS Code extension)](https://marketplace.visualstudio.com/items?itemName=Anthropic.claude-code)
* Any other MCP-compatible application

## Troubleshooting

<AccordionGroup>
  <Accordion title="MCP server not appearing in client">
    * Verify the configuration file path is correct for your operating system
    * Ensure the JSON syntax is valid
    * Restart your MCP client after making changes
    * Check that `m87` is accessible (run `which m87` or use absolute path)
  </Accordion>

  <Accordion title="Authentication errors">
    Make sure you're logged in:

    ```bash theme={null}
    m87 login
    ```

    The MCP server uses your existing m87 credentials.
  </Accordion>

  <Accordion title="Command not found errors">
    If you see errors about `m87` not being found:

    1. Find the full path: `which m87`
    2. Use the absolute path in your configuration:

    ```json theme={null}
    {
      "command": "/full/path/to/m87",
      "args": ["mcp"]
    }
    ```
  </Accordion>

  <Accordion title="Testing the MCP server">
    You can test the MCP server directly:

    ```bash theme={null}
    m87 mcp
    ```

    This will start the server in stdio mode. Press Ctrl+C to exit.
  </Accordion>
</AccordionGroup>

## Security Considerations

* The MCP server uses your existing m87 authentication
* All operations are subject to your device permissions
* Audit logs track all device interactions, including those via MCP
* The MCP server runs with the same privileges as your user account

## Learn More

* [Model Context Protocol specification](https://modelcontextprotocol.io)
* [m87 CLI Reference](/api/cli-reference)
* [MCP Tools Reference](/api/mcp-tools)