device_activity
ChatGPTWhich vehicles ran during a period and which never did, with each one's last engine-on moment and assigned driver. Use it for "which vehicles have not moved this week?", "what is sitting idle?", "is anything not reporting?", or a utilisation check before hiring or selling. Ask for status "inactive" to get only the vehicles that never ran in the period -- that is the question this answers best. Period is at most 31 days.
find_vehicle
ChatGPTFind a vehicle the current user has access to by its name or license plate, then return its latest known location: street address, coordinates, its current status (the same status the user sees on the tracking website, with the speed when it is moving), and the time it last reported. Use this to answer questions such as "where is <vehicle>?" or "is <vehicle> moving?".
fuel_consumption
ChatGPTReport how much fuel was actually consumed over a date range (up to 31 days), for the whole fleet or one vehicle: fuel burned, fuel burned while idling, distance driven, and average fuel economy, with a per-vehicle breakdown ranked by usage. Use this for questions like "how much fuel did we use this month?" or "what was <vehicle>'s fuel consumption last week?". This is fuel BURNED (measured from trips), not fuel purchased; for fuel-card spend use list_refuels instead. Every measurement is reported in the account's own units, and each value is written with the unit it is in. Note that fuel economy is reported either as litres per 100km, which falls as efficiency rises, or as mpg, which rises.
list_alerts
ChatGPTList the alerts that fired for the user's vehicles over a date range of up to 3 days: speeding, entering or leaving a geofence, temperature out of range, long idling, lost connection, harsh braking and so on. Each alert says what happened, when, where, and which vehicle. Use this to answer "did anything happen last night?", "has anything gone wrong with <vehicle>?", or "why did I get a notification?". This reports only alerts the user has already configured; it cannot create or change them. A vehicle with no alerts set up will never appear here, so an empty result means nothing was configured or nothing fired, not that the vehicle behaved.
list_events
ChatGPTShow a single vehicle's activity timeline over a date range (up to 31 days): its drives and stops (parkings) in order, with totals for driving time, parked time, distance, and top speed. Use this to answer "what did <vehicle> do that day?", "how long was <vehicle> parked?", or to see where and how long a vehicle stopped. Identify the vehicle by name or plate. For just the trips (no stops), use list_trips instead. Every measurement is reported in the account's own units, and each value is written with the unit it is in.
list_penalties
ChatGPTList traffic penalties (fines) issued against the user's vehicles, from the protocols.ge feed: what the fine was for, when, how much, whether it has been paid, and links to the photo or video evidence. Optionally restrict to one vehicle, a date range, or only unpaid fines. Use this to answer "do I have any unpaid fines?", "how much did we pay in fines last month?", or "why was <vehicle> fined?". Amounts are reported in lari. Evidence links expire about 2 hours after this answer, so tell the user to open them promptly rather than saving them for later.
list_refuels
ChatGPTSummarize fuel-card refuels (station purchases recorded via fuel cards) for the current user's vehicles over a date range: how many purchases, total litres, and total cost, with a per-vehicle breakdown. Optionally restrict to a single vehicle by name or license plate. Use this for questions about fuel-card spend and purchase counts, e.g. "how much did we spend on fuel this month?". Note: this reflects card purchases, not tank-sensor-detected fillings.
list_reports
ChatGPTList the reports run_report can run, with the full description of each: the columns it returns, what it needs, and what it is for. Call this when run_report's own summary list is not enough to choose between two reports, when a report name you tried was rejected, or when the caller asks what reports are available. Pass "about" with a word or two to narrow the list ("fuel", "driver", "temperature", "geofence"); omit it to get all of them, grouped. This only describes reports. It does not run one and it reads no fleet data, so it costs nothing to call before deciding.
list_trips
ChatGPTList the trips a single vehicle made over a date range (up to 31 days): each trip's start and end time, where it started and ended, distance, duration, and top speed, plus totals for the period. Use this to answer "where has <vehicle> been?", "how far did <vehicle> drive last week?", or to review a vehicle's activity. Identify the vehicle by name or plate. Every measurement is reported in the account's own units, and each value is written with the unit it is in.
list_vehicles
ChatGPTList the vehicles the current user has access to, with each vehicle's current status and a summary count. Optionally restrict to a single vehicle group by name. Use this to answer "what vehicles do I have?", "how many are moving right now?", or as an overview before drilling into a specific vehicle. Each vehicle reports the same status the user sees on the tracking website, along with the time it last reported.
maintenance_due
ChatGPTList the servicing and maintenance a user's vehicles are due for: what is overdue, what is coming up, and what is merely booked. Each item names the vehicle, the kind of service, its state and its cost. Use this to answer "does anything need servicing?", "what is overdue?", or "when was <vehicle> last serviced?". This reports the state the system has already worked out for each service, which is what the user sees in their own dashboard. It reports schedules; it cannot create, complete or change one.
nearest_objects
ChatGPTWhich of your company's saved objects (customer sites, depots, points of interest) are closest to a place, nearest first. Use it for "what is near this vehicle?", "which site is closest to <place>?", or to find the name of a location you only have coordinates for. The place can be another saved object by name, or coordinates as "lat,lon". To ask what is near a vehicle, look the vehicle up first and pass its coordinates.
nearest_vehicles
ChatGPTWhich of your vehicles are closest to a place, right now, nearest first. Use it for "who is nearest the depot?", "which van should take this job?", or "is anyone near <site>?". The place can be one of your own saved objects by name, or coordinates as "lat,lon". Optionally keep only vehicles with the engine running or only those stopped. Positions are the same live snapshot the tracking map shows.
objects_serviced
ChatGPTReport which of your company's objects (customer sites, depots, and other saved locations) a vehicle stopped at over a date range (up to 31 days), and how many times. Use this to answer "how many objects did <vehicle> service last week?", "which sites did we visit?", or "how many customers did this vehicle call on?". Covers the whole fleet by default, or one vehicle by name or plate. A stop counts as a visit when the vehicle stopped inside the object's own radius; where several objects overlap, the stop is credited only to the nearest one, so a visit count is always a count of real stops.
run_report
ChatGPTRun one of the system's own reports and return its table. Reach for this when the question maps to a named report, when the caller asks for "the <x> report", or when they want a report's own columns rather than a prose answer. For a plain question ("where is it", "how much fuel last month") the dedicated tools above answer better, because they answer in sentences. Pick the report from this list and pass its key as "report". Each line is the question that report answers. If two look close, the difference is named on the line; if you are still unsure, or the caller asks what reports exist, call list_reports first rather than guessing. TRIPS, STOPS AND UTILISATION startstop: Every trip with its start and end place and time. The default driving log. startstopfull: The same trips with fuel and work time added per trip. stops: Every stop: where it stood, when, and for how long. stationary_work: Standing still with equipment running (crane, pump, PTO), not driving. period: Usage totals per vehicle, or day by day when one vehicle is named. common: Fleet totals for the period. group_by device, days or devicebyday. activitybyday: One row per vehicle per day, first start to last stop. workingperiod: Engine-on, driving and idle time. The utilisation and idling answer. workingperiod2: Per work day with full fuel accounting, including hand corrections. workingperiod3: Per day from first to last transmission, with fuel. firststart: When and where each vehicle first moved. One day only. odometer: Recorded odometer readings against tracked distance, for one month. transmissions_turn_off: When a device or ignition was switched on and off, and where. tachometer_ranges: How long the engine spent in each RPM band. DRIVERS driverroutes: Per driver: trips, distance, move time, top speed and fuel. devicedrivers: Which drivers drove which vehicle, and how far under each. drivedevices: Which vehicles one driver used. Keyed by driver, not by vehicle. Needs driver. driver_workhours: One row per trip inside a driver's shift window. Needs driver. driver_workhours_by_day: A driver's day totals plus ecodrive event counts. Needs driver. drivercheckin: Check-in and check-out at sites, with photos and time spent. drivertasks: Assigned tasks and whether the driver completed them. driver_images: Photos drivers sent in chat over the period. driver_object_visits: Whether a planned route's objects were visited. Needs route. FUEL fuel: Fuel burned against each vehicle's configured budget. fuelgeneral: Per-vehicle fuel picture: economy, tank level, refills. Takes fuel_source. fuelperiod: One vehicle's fuel level over time, refuels and drains. Needs vehicle and fuel_source. fuelfilling: Fuel-card refills: station, litres, unit price and cost. fuelssb: Expected against actual fuel cost per day, and the deviation. Takes fuel_source. ALERTS AND DRIVING BEHAVIOUR alerts: Every alert that fired, each with a plain-text description. alerts_speedlimit: Speeding only: top speed, the limit, and how long it was held. alerts_camera: Camera events: fatigue, phone use, seatbelt. Needs vehicle. ecodrive_event: Individual harsh braking, acceleration, turning and idling events. ecodrive_europe: A risk score and zone per vehicle, or per driver via entry_type. PLACES, OBJECTS AND GEOFENCES stopnearobject: Stops inside one of your objects, with driver, address and dwell time. stopnearobject_ranges: The same question, compact: which vehicle stopped where, and how long. geofences_by_device_detailed: One vehicle's geofence entries, exits and time inside. Needs vehicle. checkpointsstartstop: Checkpoint-to-checkpoint legs with temperature and fuel. Cold chain. TEMPERATURE AND WIRED SENSORS temperature: The raw probe trace, ping by ping. Needs vehicle, at most 7 days. temperature_ranges: Time spent in each temperature band. Proves a load stayed in range. sensor_state: When a wired input (door, pump, PTO) switched on or off. Needs vehicle. ENGINE DIAGNOSTICS obd_errorcods: Dashboard warning lights the vehicle reported. Needs vehicle. obd_rpmranges: RPM bands with time and fuel burned in ea…
temperature_readings
ChatGPTReport what a vehicle's temperature probes actually recorded over a date range (up to 31 days): the lowest, highest and average reading per probe, when the extremes happened, and how long the vehicle held a range you give. Use this for "did the load stay cold?", "how cold did the trailer get last week?", or to check a refrigerated or pharmaceutical delivery. Covers one vehicle at a time, named by name or plate, and reports every probe fitted to it: a truck's probes are different compartments, so a chilled box and a warm cabin on the same vehicle are both normal. This is different from list_alerts. list_alerts reports that a limit someone configured was breached and fired. This reports what the probes read, whether or not anyone configured a limit, so it can show a load riding just inside its range all week. Temperatures are reported in the account's own unit, and each value is written with the unit it is in. Give "min" and "max" in that same unit. An empty answer means the vehicle has no temperature probes fitted, not that it was never cold.
vehicle_details
ChatGPTReturn reference details for a single vehicle the current user has access to, found by name or license plate: manufacturer, model, year, VIN, fuel type, tank size, odometer, and the assigned driver(s). Use this to answer "what do we know about <vehicle>?" or questions about a vehicle's specs. For its live position, use find_vehicle instead.
vehicle_search_by_place
ChatGPTFind which vehicles were at a place during a period, and when. Use it for "was anyone at <address> last Tuesday?", "which vehicle visited this location?", or to check a disputed delivery or callout. The place can be one of your saved objects by name, or coordinates as "lat,lon"; the radius is in metres. By default it searches every position the vehicle reported while moving; set only_stops to true to find only where a vehicle actually came to a stop, which is what answers "did anyone stop here". Period is at most 31 days.
working_hours
ChatGPTReport how long vehicles actually worked over a date range (up to 31 days): engine-on time, of which driving time and idling time (engine running but not moving), plus distance. Use this for "how many hours did <vehicle> work last week?", "which vehicles are idling too much?", or to compare vehicle utilisation. Covers the whole fleet by default, or one vehicle by name or plate. Times are hours:minutes:seconds. Distances are reported in the account's own unit, and each value is written with the unit it is in.