NewThe Trackpac MCP server

Your sensors,now fluent in AI.

Connect Claude, ChatGPT and other AI assistants to your Trackpac sensors, sites and alerts. Ask about the physical world in plain language, and set up facilities just by describing them.

Hosted at mcp.trackpac.ioNothing to installOpen source

Which freezers went above −18 °C last night?
TRACKPACMCP
Temperature and humidity sensor
Indoor air quality monitor
Water leak sensor
Tank level sensor
GPS asset tracker
Soil moisture sensor

Just ask. Like you would a colleague.

  • Which sensors have a low battery?
  • Show CO₂ in every meeting room this week
  • Alert me if the server room goes above 27 °C
  • Where is trailer 12 right now?
  • Create a geofence around the Leeds warehouse
  • Which tanks are below 20%?
  • Build a dashboard for Building A
  • Set every freezer sensor to report every 10 minutes
  • Email a weekly temperature report to facilities
  • Has any leak sensor triggered today?
  • Which devices haven't reported in 24 hours?
  • Move these 14 sensors to the Rotterdam site
  • Compare humidity across all three greenhouses
  • Who gets notified when the cold room alarms?

How it works

From question to action, in seconds.

The Trackpac MCP server gives your AI assistant around 130 tools covering the Trackpac API. It picks the right ones, calls them, and tells you what it found or did.

01 · Monitor

Ask about anything your sensors can see.

Your assistant finds the right devices, pulls their readings and history from Trackpac, and answers in plain language, with the numbers to back it up.

Assistant · Trackpac connected

02 · Set up

Stand up a new facility in a single sentence.

Describe the site you want. Your assistant draws the geofence, tags the sensors and wires up the alerts, asking you to confirm before it changes anything.

Assistant · Trackpac connected

03 · Build

Get a dashboard without building a dashboard.

Ask for the view you need. Your assistant picks the devices, adds the widgets, lays them out and shares a link with your team.

Assistant · Trackpac connected

Illustrative examples. Your assistant answers from your own Trackpac data.

Sense

Every sensor in your estate. One conversation.

Temperature, humidity, air quality, leaks, tank levels, soil moisture and location. If it reports into Trackpac, your AI can read it, chart it and act on it.

Leak detection monitoring

Water leak

Leak detection

“Has any leak sensor triggered today?”

Cold storage monitoring

Temperature

Cold storage

“Which fridges ran warm overnight?”

Server rooms monitoring

Temperature & humidity

Server rooms

“Alert me above 27 °C.”

Greenhouses monitoring

Humidity

Greenhouses

“Compare all three greenhouses.”

Tanks monitoring

Fill level

Tanks

“Which tanks are below 20%?”

Fields & vineyards monitoring

Soil moisture

Fields & vineyards

“Where do we need to irrigate?”

Works with the LoRaWAN hardware Trackpac already supports, from Dragino, Milesight and Browan to RAKwireless, Tekelek and Digital Matter.

  • Milesight EM300-TH

    Temperature & humidity

    Milesight EM300-TH

  • Milesight AM319

    Indoor air quality

    Milesight AM319

  • Milesight water leak sensor

    Water leak

    Milesight water leak sensor

  • Tekelek TEK766

    Tank level

    Tekelek TEK766

  • Digital Matter Yabby3

    GPS location

    Digital Matter Yabby3

  • Dragino SE01-LB

    Soil moisture

    Dragino SE01-LB

  • MerryIoT air quality sensor

    CO₂ & air quality

    MerryIoT air quality sensor

  • RAKwireless RAK2270

    Cold chain tracking

    RAKwireless RAK2270

  • Dragino LDDS20

    Liquid level

    Dragino LDDS20

  • Browan TBOL100

    Asset tracking

    Browan TBOL100

  • Dragino LWL02

    Leak detection

    Dragino LWL02

  • Browan TBWL100

    Water leak

    Browan TBWL100

Create & manage

Facilities, managed by conversation.

Your AI doesn't just read sensor data. With your approval, it can create and change things in Trackpac: sites, geofences, dashboards, assets and alerts.

app.trackpac.io
Creating a geofence on a 3D map in Trackpac

“Put a 250 m geofence around our Rotterdam depot and tell ops when a trailer leaves.”

Geofences

Draw a site with a sentence.

Geofences from a plain description. Your assistant looks up the address, sets the radius, and decides who hears about it when something enters or leaves.

perform_geocoding_location_engine_geocode_post
create_geofence_geofence_post
activate_geofence_geofence
app.trackpac.io
A Trackpac sensor dashboard with temperature and humidity gauges

“Build a cold room dashboard with temperature, humidity and battery for every sensor.”

Dashboards

Dashboards that build themselves.

Describe the view you need and get a live Trackpac dashboard: the right devices, the right widgets, laid out and ready to share.

create_dashboard_dashboard
add_widget_to_dashboard_dashboard
update_dashboard_layout_dashboard
app.trackpac.io
Tracking a device's route on a 3D map in Trackpac

“Which trailers left a site today, and where are they now?”

Assets & tracking

Know where everything is.

Ask where an asset is, where it has been, or which ones left a site today. Your assistant reads location history and geofence presence and gives you the short version.

read_assets_asset_organisation
get_geofence_presence_history_geofence
read_frames_asset_history
The Trackpac mobile app
TrackpacnowFreezer B2 above −18 °CNow −14.2 °C. On-call engineer notified by SMS.

“Text the on-call engineer if any freezer goes above −18 °C.”

Alerts & actions

Alerts that set themselves up.

Thresholds, contacts and escalation in one request. Trackpac does the watching around the clock and notifies the right people by app, email or SMS.

create_contact_contact_post
create_action_action
assign_action_action_assign_post

0+

MCP tools

covering the Trackpac API

0

API areas

from devices to billing

0

Server URL

hosted by Trackpac

0

Things to install

for Claude.ai and ChatGPT

Capabilities

If Trackpac can do it, your AI can too.

The server is generated from Trackpac's live API spec, so new features show up as new tools automatically.

Devices

List, find, rename, tag, move and claim sensors across organisations.

read_devices_device_get

Live & historic data

Latest readings and full history for any device, by day or time range.

read_frames_frame

Organisations

Manage organisations, invite members and set their roles.

invite_user_organisation

Geofences

Draw zones, see who's inside, and review presence history.

create_geofence_geofence_post

Assets

Track trailers, tools and equipment and the devices attached to them.

read_assets_asset_organisation

Dashboards

Build, lay out and share dashboards and widgets.

add_widget_to_dashboard_dashboard

Alerts & actions

Set thresholds, choose contacts and read the alert log.

create_action_action

Reports

Create scheduled reports and fetch the generated files.

create_report_report

Downlinks

Reconfigure devices over the air, like changing reporting intervals.

send_downlink_device

Location engine

Geocode addresses and resolve positions from LoRaWAN and Wi-Fi.

resolve_location_location_engine_resolve_post

Works with the AI you already use.

Built on the open Model Context Protocol, so it isn't tied to any one assistant. Every client connects to the same hosted server.

Claude.aiSign in
ChatGPTSign in
Claude DesktopAPI key
Claude CodeAPI key
VS Code & CopilotOne-click install
CursorOne-click install
CodexAPI key
Any MCP clientStreamable HTTP

Get started

Connected in minutes. Nothing to install.

Trackpac hosts the MCP server for you. Point your AI assistant at it, then sign in with your Trackpac account or use an API key.

  1. 1

    Get an API key

    Using Claude.ai or ChatGPT? Skip this step. You'll sign in with your Trackpac account instead.

    1. Open your Trackpac account page.
    2. Click Generate API key.
    3. Give it a name, like “Claude Desktop – laptop”.
    4. Copy the key and keep it somewhere safe.
    Open app.trackpac.io
  2. 2

    Connect your assistant

    Every client connects to the same hosted server.

    SERVER URL

    https://mcp.trackpac.io/mcp
    Sign in with Trackpac
    Claude.ai setup
    1. In Claude, open Customize → Connectors → Add custom connector.
    2. Name it Trackpac and paste in the server URL.
    3. Choose Sign in now and log in with your Trackpac account.
    4. In a chat, turn on Trackpac under + → Connectors.

    Works on claude.ai and in Claude Desktop. On Team or Enterprise plans, an owner adds it in Organization settings → Connectors. Full Claude.ai guide →

Then just ask.

Try “List my Trackpac devices” to check everything is working.

Prefer to run it yourself?

The server is open source under Apache 2.0. Run it locally or with Docker. Self-hosting guide →

Security

Powerful, and still in your hands.

The MCP server is a thin layer over the Trackpac API, and it's open source under Apache 2.0. Your data stays in Trackpac, and every request runs with your own Trackpac permissions.

Your permissions, nothing more

The assistant acts as you. It can only see and change what your Trackpac user already can.

Approve before it acts

Claude and ChatGPT ask before running tools that change things. Keep those prompts on for anything that writes or deletes.

Sign-in or keys you control

Sign in with your Trackpac account, or create a separate API key for each machine and revoke any of them from your account page.

Give your AI a sense of the physical world.