AI-Powered UniFi Management
MCP servers that give AI agents direct access to your UniFi infrastructure. Query, analyze, and manage your network — safely and intelligently.
Servers
One monorepo, multiple MCP servers — each tailored to a UniFi application.
Network
Full UniFi Network controller coverage. Devices, clients, firewall policies, WLANs, port forwarding, VLANs, DPI stats, and more.
Explore Network →Protect
UniFi Protect camera management. Live snapshots, event timelines, motion detection, smart alerts, and recording access.
Explore Protect →Access
Door access control, visitor management, and credential provisioning for UniFi Access hubs, locks, and readers.
Drive
Network-attached storage management for UniFi Drive. File operations, storage monitoring, and backup automation.
Quick Install
Get up and running in under a minute with your preferred method.
# Network server
uvx unifi-network-mcp
# Protect server
uvx unifi-protect-mcpdocker run -i --rm \
-e UNIFI_NETWORK_HOST=192.168.1.1 \
-e UNIFI_NETWORK_USERNAME=admin \
-e UNIFI_NETWORK_PASSWORD=your-password \
ghcr.io/sirkirby/unifi-network-mcp:latest{
"mcpServers": {
"unifi-network": {
"command": "uvx",
"args": ["unifi-network-mcp"],
"env": {
"UNIFI_NETWORK_HOST": "192.168.1.1",
"UNIFI_NETWORK_USERNAME": "admin",
"UNIFI_NETWORK_PASSWORD": "your-password"
}
}
}
}Why UniFi MCP?
Built for AI agents. Designed for safety. Optimized for LLM efficiency.
125+ Tools
Comprehensive coverage across Network and Protect controllers. Devices, clients, firewall, cameras, events, and more.
Safe by Default
Read-only by default. Mutations require explicit opt-in and a preview-then-confirm flow before any changes hit your controller.
Context Optimized
Lazy tool loading uses ~200 tokens vs ~5,000 for eager mode. Built for LLM efficiency from the ground up.
Multi-Transport
stdio, Streamable HTTP, and SSE. Run locally with Claude Desktop or expose remotely for automation platforms.
Docker Ready
Pre-built containers for both servers with a docker-compose file included. Deploy in minutes.
Open Source
MIT licensed and community-driven. Contributions welcome — from new tools to entire server implementations.