Fleet MCP

Fleet MCP

A Model Context Protocol (MCP) server that enables AI assistants to interact with Fleet Device Management for device management, security monitoring, and compliance enforcement.

Demo

📸 Show/Hide Demo Screenshots

Features

• Host Management: List, search, query, and manage hosts across your fleet
• Live Query Execution: Run osquery queries in real-time against hosts
• Policy Management: Create, update, and monitor compliance policies
• Software Inventory: Track software installations and vulnerabilities across devices
• Team & User Management: Organize hosts and users into teams
• Osquery Table Discovery: Dynamic discovery and documentation of osquery tables
• Read-Only Mode: Safe exploration with optional SELECT-only query execution
• Activity Monitoring: Track Fleet activities and audit logs

Quick Start

Just want to dive right in? This will set up fleet-mcp with read-only access and SELECT query execution enabled. Just replace the FLEET_SERVER_URL and FLEET_API_TOKEN with your own.

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

See the Available Tools section below for a complete list of tools.

---

<!--

Local Installation

From PyPI

pip install fleet-mcp

From Source

git clone https://github.com/SimplyMinimal/fleet-mcp.git

cd fleet-mcp pip install -e .

Using uv (recommended for development)

git clone https://github.com/SimplyMinimal/fleet-mcp.git

cd fleet-mcp uv sync --dev

</details> -->

<!-- ### 1. Initialize Configuration

fleet-mcp init-config

This creates a fleet-mcp.toml configuration file. Edit it with your Fleet server details:

[fleet]

server_url = "https://your-fleet-instance.com" api_token = "your-api-token" readonly = true # Safe default - enables read-only mode allow_select_queries = false # Set to true to allow SELECT queries

2. Test Connection

fleet-mcp test

3. Run the MCP Server

fleet-mcp run

MCP Client Configuration

Fleet MCP can be integrated with various MCP-compatible clients. Below are configuration examples for popular clients.

Prerequisites

Before configuring any MCP client, ensure you have:

1. Install `uv` (recommended) or pip:

   # Install uv (recommended)

curl -LsSf https://astral.sh/uv/install.sh | sh

Or use pip

pip install fleet-mcp

1. Fleet API Token: Generate an API token from your Fleet instance: Option 1)

• Log into Fleet UI
• Navigate to: My account → Get API token
• Copy the token for use in configuration

Option 2)

• Create an API-Only user with fleetctl

   # Generate an API-Only User and get the token

fleetctl user create --name Fleet-MCP --email --password --role admin --api-only

Note: This API token and your fleet instance URL (https://your-fleet-instance.com) will be used in the client configuration.

1. Pick Your Client: Choose your preferred AI assistant client and follow the corresponding setup instructions below.

Install in Claude Desktop

Configuration File Location

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

Configuration

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Note: Replace uvx with fleet-mcp if you've installed the package globally. For enhanced security, use --config flag to reference a TOML file instead of embedding tokens (see Security Best Practices).

Install in Cursor

Go to: Settings → Cursor Settings → MCP → Add new global MCP server

Install globally in ~/.cursor/mcp.json or per-project in .cursor/mcp.json. See Cursor MCP docs for more info.

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Install in Cline (VS Code Extension)

Config Location: ~/.cline/mcp_settings.json (macOS/Linux) or %USERPROFILE%\.cline\mcp_settings.json (Windows)

Alternatively: VS Code Settings → Search "Cline: MCP Settings" → Edit JSON

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Install in Continue (VS Code Extension)

Config Location: ~/.continue/config.json

{

"mcpServers": [ { "name": "fleet", "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } ] }

Install in Zed Editor

Config Location: ~/.config/zed/settings.json (macOS/Linux) or %APPDATA%\Zed\settings.json (Windows)

{

"context_servers": { "fleet": { "command": { "path": "uvx", "args": ["fleet-mcp", "run"] }, "settings": { "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } } }

Install in Windsurf

See Windsurf MCP docs for more info.

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Install in VS Code

See VS Code MCP docs for more info.

"mcp": {

"servers": { "fleet": { "type": "stdio", "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Install in Sourcegraph Cody

Config Location: ~/Library/Application Support/Cody/mcp_settings.json (macOS), %APPDATA%\Cody\mcp_settings.json (Windows), or ~/.config/Cody/mcp_settings.json (Linux)

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Install in Augment Code

Via UI: Hamburger menu → Settings → Tools → + Add MCP → Enter uvx fleet-mcp run → Name: "Fleet" → Add

Manual Config: Settings → Advanced → Edit settings.json

"augment.advanced": {

"mcpServers": [ { "name": "fleet", "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } ] }

Install in LM Studio

Navigate to Program → Install → Edit mcp.json. See LM Studio MCP Support.

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Generic MCP Client Configuration

For other MCP-compatible clients, use this general pattern:

{

"mcpServers": { "fleet": { "command": "uvx", "args": ["fleet-mcp", "run"], "env": { "FLEET_SERVER_URL": "https://your-fleet-instance.com", "FLEET_API_TOKEN": "your-api-token", "FLEET_READONLY": "true", "FLEET_ALLOW_SELECT_QUERIES": "true" } } } }

Configuration Options Reference

Note: All clients above use the same environment variables. Replace uvx with fleet-mcp if installed globally.

Security Best Practices

1. Use Config Files: Store tokens in TOML files: "args": ["fleet-mcp", "--config", "~/.config/fleet-mcp.toml", "run"]
2. File Permissions: chmod 600 ~/.config/fleet-mcp.toml
3. Read-Only Mode: Start with FLEET_READONLY=true (default)
4. Token Rotation: Regularly rotate Fleet API tokens
5. Environment-Specific Configs: Use s