Manage multiple UniFi sites from one AI assistant

If you manage more than one UniFi site—whether for a business with multiple offices, a managed service provider with many customers, or a home lab with separate network segments—the UniFi MCP Server lets your AI assistant work across all of them from a single session. You can list sites, compare performance metrics, monitor health scores, and aggregate device and client statistics without switching between consoles.

Multi-site management tools require the Site Manager API. Set UNIFI_SITE_MANAGER_ENABLED=true and provide your SITE_MANAGER_API_KEY from api.ui.com. Without this, most tools in this section will return an error.

What you can do

List all sites

Retrieve all sites with aggregate device counts, client counts, and bandwidth stats.

Health monitoring

Get health scores and status for all sites—healthy, degraded, or down.

Cross-site analytics

Aggregate totals and compare performance metrics across every site you manage.

Cross-site search

Find a device, client, or network across all sites in one query.

The `site_id` parameter

Every tool in the UniFi MCP Server that operates on a specific site accepts a site_id parameter. When you have multiple sites, pass the appropriate site ID to scope any operation to that location.

# List devices on the New York office site
devices = await mcp.call_tool("list_devices_by_type", {
    "site_id": "nyc-office",
    "device_type": "uap"
})

# Get network stats for the London office
stats = await mcp.call_tool("get_network_statistics", {
    "site_id": "lon-office"
})

Use list_all_sites (read below) to discover the site IDs for your deployment.

Listing all sites

Example prompts:

“Show me all my UniFi sites”
“How many sites are there in total?”
“List all sites with their device counts”

sites = await mcp.call_tool("list_all_sites_aggregated", {})

Response:

[
  {
    "site_id": "nyc-office",
    "name": "New York Office",
    "device_count": 12,
    "client_count": 85,
    "status": "healthy"
  },
  {
    "site_id": "lon-office",
    "name": "London Office",
    "device_count": 8,
    "client_count": 42,
    "status": "healthy"
  },
  {
    "site_id": "chi-branch",
    "name": "Chicago Branch",
    "device_count": 4,
    "client_count": 18,
    "status": "degraded"
  }
]

Site health monitoring

The get_site_health_summary tool returns a health summary that includes device and client counts, uptime percentage, and an overall status for each site. Example prompts:

“Which sites are having problems?”
“Show the health score for the London office”
“Are all my sites healthy?”

# Health summary for all sites
summaries = await mcp.call_tool("get_site_health_summary", {})

# Health summary for a specific site
summary = await mcp.call_tool("get_site_health_summary", {
    "site_id": "chi-branch"
})

Response (single site):

{
  "site_id": "chi-branch",
  "site_name": "Chicago Branch",
  "status": "degraded",
  "devices_total": 4,
  "devices_online": 3,
  "clients_active": 18,
  "uptime_percentage": 75.0
}

Health statuses:

Status	Meaning
`healthy`	All devices online, no issues detected
`degraded`	Some devices offline or performance issues
`down`	Site unreachable or critical failure

Cross-site aggregate statistics

get_cross_site_statistics aggregates device counts, client counts, bandwidth, and health status across every site you manage in a single call. Example prompts:

“Give me a summary across all my sites”
“How many total devices and clients do I manage?”

stats = await mcp.call_tool("get_cross_site_statistics", {})

Response:

{
  "total_sites": 3,
  "sites_healthy": 2,
  "sites_degraded": 1,
  "sites_down": 0,
  "total_devices": 24,
  "devices_online": 23,
  "total_clients": 145,
  "total_bandwidth_up_mbps": 0.0,
  "total_bandwidth_down_mbps": 0.0,
  "site_summaries": [...]
}

Comparing site performance

compare_site_performance ranks all sites by uptime and device availability, identifying your best and worst performers. Example prompts:

“Which site is performing the worst this week?”
“Compare network performance across all my offices”
“Which location has the highest uptime?”

comparison = await mcp.call_tool("compare_site_performance", {})

Response:

{
  "total_sites": 3,
  "best_performing_site": {
    "site_id": "nyc-office",
    "site_name": "New York Office",
    "uptime_percentage": 99.9,
    "device_online_percentage": 100.0,
    "avg_latency_ms": 12.5,
    "health_status": "healthy"
  },
  "worst_performing_site": {
    "site_id": "chi-branch",
    "site_name": "Chicago Branch",
    "uptime_percentage": 75.0,
    "device_online_percentage": 75.0,
    "avg_latency_ms": 45.2,
    "health_status": "degraded"
  },
  "average_uptime": 91.6,
  "average_latency_ms": 24.0,
  "site_metrics": [...]
}

Searching across all sites

search_across_sites lets you find a device, client, or network across every site without knowing which site it belongs to. Example prompts:

“Find the device named ‘Server-Room-AP’ across all my sites”
“Which site is the client with MAC aa:bb:cc:dd:ee:ff on?”
“Search for the ‘Guest’ network across all locations”

result = await mcp.call_tool("search_across_sites", {
    "query": "server-room",
    "search_type": "device"
})

Response:

{
  "total_results": 2,
  "search_query": "server-room",
  "result_type": "device",
  "results": [
    {
      "type": "device",
      "site_id": "nyc-office",
      "site_name": "New York Office",
      "resource": {
        "name": "Server-Room-AP",
        "mac": "aa:bb:cc:dd:ee:ff",
        "model": "U6-Pro"
      }
    }
  ]
}

Set search_type to device, client, network, or all.

Site inventory

get_site_inventory provides a detailed breakdown of resources at one or all sites—device counts by type, client counts, network counts, SSID counts, VPN tunnels, and firewall rules.

# Inventory for a specific site
inventory = await mcp.call_tool("get_site_inventory", {
    "site_id": "nyc-office"
})

# Inventory for all sites
all_inventories = await mcp.call_tool("get_site_inventory", {})

Response:

{
  "site_id": "nyc-office",
  "site_name": "New York Office",
  "device_count": 12,
  "device_types": {
    "uap": 8,
    "usw": 3,
    "ugw": 1
  },
  "client_count": 85,
  "network_count": 5,
  "ssid_count": 3,
  "uplink_count": 2,
  "vpn_tunnel_count": 1,
  "firewall_rule_count": 14
}

Internet health and ISP metrics

Monitor ISP performance across sites with latency, bandwidth, jitter, and packet loss data. Example prompts:

“What is the internet health for the Chicago office?”
“Show latency and packet loss for all sites”

# Internet health across all sites
health = await mcp.call_tool("get_internet_health", {})

# ISP metrics for a specific site
isp = await mcp.call_tool("get_isp_metrics", {
    "site_id": "chi-branch"
})

Response:

{
  "site_id": "chi-branch",
  "isp_name": "Example ISP",
  "download_bandwidth_mbps": 480.0,
  "upload_bandwidth_mbps": 95.0,
  "latency_ms": 45.2,
  "jitter_ms": 8.1,
  "packet_loss_percent": 0.5,
  "timestamp": "2026-04-22T12:00:00Z"
}

Managed hosts

If you manage UniFi consoles (UDM Pro, Cloud Key, etc.), you can list all hosts and check their status.

hosts = await mcp.call_tool("list_hosts", {
    "limit": 50,
    "offset": 0
})

Response:

[
  {
    "host_id": "host-001",
    "hostname": "udm-pro-nyc",
    "ip_address": "192.168.1.1",
    "model": "UDM-Pro",
    "version": "3.2.9",
    "site_count": 1,
    "status": "online",
    "last_seen": "2026-04-22T12:00:00Z"
  }
]

Example: morning site health check

This workflow shows how to use multi-site tools for a daily operational review.

Get aggregate status

stats = await mcp.call_tool("get_cross_site_statistics", {})
# Check: sites_degraded or sites_down > 0

Find problem sites

summaries = await mcp.call_tool("get_site_health_summary", {})
# Filter for status == "degraded" or "down"

Drill into the worst site

isp = await mcp.call_tool("get_isp_metrics", {
    "site_id": "chi-branch"
})
# Check latency, packet loss, bandwidth

Compare all sites

comparison = await mcp.call_tool("compare_site_performance", {})
# Review best and worst performers

Tool availability by API mode

Tool	Local API	Cloud EA	Cloud V1
`list_all_sites_aggregated`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`get_site_health_summary`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`get_cross_site_statistics`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`compare_site_performance`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`search_across_sites`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`get_site_inventory`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`get_isp_metrics`	Requires Site Manager	Requires Site Manager	Requires Site Manager
`list_hosts`	Requires Site Manager	Requires Site Manager	Requires Site Manager

All multi-site tools are read-only. They query and aggregate data across sites but do not modify any configuration. No confirm=True is required for any multi-site tool.

Getting Started

Installation

MCP Client Setup

Features

Advanced

Manage multiple UniFi sites from one AI assistant

What you can do

List all sites

Health monitoring

Cross-site analytics

Cross-site search

The `site_id` parameter

Listing all sites

Site health monitoring

Cross-site aggregate statistics

Comparing site performance

Searching across all sites

Site inventory

Internet health and ISP metrics

Managed hosts

Example: morning site health check

Tool availability by API mode

​What you can do

List all sites

Health monitoring

Cross-site analytics

Cross-site search

​The site_id parameter

​Listing all sites

​Site health monitoring

​Cross-site aggregate statistics

​Comparing site performance

​Searching across all sites

​Site inventory

​Internet health and ISP metrics

​Managed hosts

​Example: morning site health check

​Tool availability by API mode

What you can do

The `site_id` parameter

Listing all sites

Site health monitoring

Cross-site aggregate statistics

Comparing site performance

Searching across all sites

Site inventory

Internet health and ISP metrics

Managed hosts

Example: morning site health check

Tool availability by API mode