Skip to content
  • There are no suggestions because the search field is empty.

What the GlacierGrid connector can do

Everything Claude can read through the connector, grouped by what you'd use it for - all of it read-only.

Companion to the connection guide.

You don't need to know any of this to use the connector - ask questions in plain language and Claude picks the right capability. This page is for people who want to know exactly what it can reach before connecting it.

Every capability below only reads. As the connector stands today, none of them can change a setpoint, acknowledge an alert, create a work order, or send a message.

If capabilities that change data are added later, they will be marked as such and Claude will ask for your confirmation before running one. That distinction is built into the protocol rather than left to us: read-only operations can run without a prompt, and anything that modifies or deletes data always prompts.

Everything is scoped to your own company, and further scoped to the locations your GlacierGrid permissions cover.

Finding things

Search - Turns a name you'd actually say ("the Dallas store", a serial number, an equipment name) into the specific record it refers to. Almost every question starts here, which is why Claude sometimes looks something up before answering.

Site overview - The full picture for one site or your whole account: equipment, sensors, thermostats, gateways, open issues, plus the site's timezone and location.

Temperature and equipment

Sensor readings - Temperature and humidity history from your monitoring sensors, for a single device, a site, or the whole account. Returned as a trend rather than a raw dump when the window is long.

Thermostat status - Live state for supported smart thermostats: zone temperature, setpoints, which heating or cooling stages are running, and whether the unit is online.

Thermostat cycle summary - How hard equipment is working over a period: how often it cycles, total runtime, how well it recovers, and whether it's delivering the temperature change it should.

Equipment HVAC status - A single combined snapshot for one zone: its configuration, thermostat, sensors, recent control activity and open issues.

Refrigeration summary - Condenses cooler and freezer telemetry into per-unit health signals. Two modes: a trend view for spotting units degrading toward failure, and a compliance view reporting monitoring coverage, time in range, and sustained excursions.

Comfort burden - How much a zone is struggling to hold its target during occupied hours, compared against its schedule and setpoints in local time.

Device connectivity - Which sensors, thermostats and gateways are online, plus signal quality, so a data gap can be told apart from an equipment problem.

Issues, alerts and activity

Issues - Active, historical or specific issues, enriched with the equipment and site they belong to and how long each has persisted.

Control events - For HVAC zones: manual overrides, scheduled adjustments, demand-response events and recovery activity, so a zone temperature change can be traced to a deliberate action. Refrigeration units have no schedules or overrides, so this doesn't apply to them.

Alert recipients - For a given alert configuration, exactly who would be notified and how they'd be reached.

Notifications sent - The record of notifications actually delivered for an issue.

Cost and context

Savings - Energy and cost savings by site and billing period, expected against actual, with a percentage you can rank sites by.

Weather - Outdoor conditions and forecasts for your sites' locations, so indoor performance can be judged against what the weather was doing.

People

Users - The people on your GlacierGrid account and their roles. Contact details are hidden by default and included only when you explicitly ask for them.

Reference material

Reference lookup - Claude can pull GlacierGrid's own diagnostic reference - the thresholds behind mechanical issue detection, how alerting behaves, how equipment states are interpreted - so its explanations match how the platform itself evaluates your equipment rather than generic HVAC advice.

How this behaves in practice

Account-wide questions stay fast. Asking about every site doesn't fetch every device one at a time - the heavier capabilities resolve and aggregate server-side, so a whole-account question is a handful of operations rather than hundreds.

Large results are marked, not silently cut. If a result is capped, the response says so, and Claude will tell you rather than presenting a partial list as complete.

Missing data is reported as missing. If a device is offline or a period has gaps, that's stated rather than filled in - a gap in monitoring is itself usually the finding.