get_device_telemetry
ChatGPTSensor readings from IoT devices over up to 24 hours: per signal the mean/min/max/last value and a series aggregated into buckets (default 15 minutes). Use for temperature, pressure, energy, vibration and other signal questions. Get device ids from list_devices.
get_failure_metrics
ChatGPTMTBF, MTTR and MTTF (hours) treating ONE stop reason as the failure event, over any period up to 90 days, total and per machine. Use when the user asks about reliability for a specific failure type (e.g. "mechanical failure"). For general maintenance KPIs use get_mtbf_mttr.
get_hourly_production
ChatGPTHour-by-hour production (good, rejected, scrap — units, weight, length) for up to 7 days, for all selected machines combined and optionally per machine (up to 5). Use to find slow hours, shift differences or when production stopped/resumed.
get_machine_status
ChatGPTReal-time status of machines right now: producing or stopped (and why, since when), connectivity, current operator, product and production order. Optionally adds today's OEE so far. Use for "what is happening now" questions.
get_machine_timeline
ChatGPTChronological states of ONE machine over up to 24 hours: producing periods and each stop with reason, sub-reason and operator note, plus totals per state. Durations count only the time inside the period. Optionally includes operator, production order and product changes. Use to explain what happened on a machine during a shift or day.
get_mtbf_mttr
ChatGPTMaintenance reliability KPIs from monthly history: MTBF (mean time between failures), MTTR (mean time to repair), MTTD (to diagnose) and MTTC (to communicate), in hours — accumulated for the period and month by month, total and per machine. Calendar periods (quarter, semester, year...) are the current block containing today.
get_oee_summary
ChatGPTOEE, availability, performance, quality, TEEP, production/downtime minutes, stop counts and good/rejected/scrap output for a period — total and per machine, optionally compared with the previous period. Periods up to 24 h use live data; longer periods use daily history per shift (up to 1 year). Use for "how did we do" questions.
get_oee_trend
ChatGPTTime series of OEE and related KPIs from daily history, grouped by day, week, month, quarter, semester or year — for all selected machines combined, per machine or per machine group. Use for trends, evolution and "best/worst week" questions.
get_production_by_product
ChatGPTProduced quantities per product over a period (good, rejected and scrap in units, weight and length), with quality, performance and production time per product, plus the total. Use for "how much did we produce" and product-mix questions.
get_production_order_metrics
ChatGPTPerformance of ONE production order (OP / job code) across all its run intervals: total duration, production and downtime, good/rejected/scrap output and performance, plus each interval per machine. Use list_production_orders to find the code.
get_spc_summary
ChatGPTStatistical process control (SPC) measurements of ONE machine over up to 7 days: per quality characteristic the number of points, mean, min, max, standard deviation, last value and a downsampled series. Use for process stability and quality-characteristic questions.
get_stop_occurrences
ChatGPTIndividual occurrences of ONE stop reason: when, on which machine, how long, sub-reason and the operator's note, plus count/average/longest statistics. Use to investigate a specific cause or read operator comments about it.
get_stop_pareto
ChatGPTPareto of stop reasons by lost time over a period: duration, occurrences, share and cumulative share per reason (or per reason group). Periods up to 24 h use live data; longer periods use daily history per shift. Use for "biggest losses / main stop causes" questions.
get_stop_reason_breakdown
ChatGPTBreaks ONE stop reason down into its sub-reasons (up to 3 levels) with lost time and number of stops each. Use after get_stop_pareto to drill into a top reason. Only works for company-specific reasons that have sub-reasons.
get_unjustified_stops
ChatGPTMost recent stops that still have no reason assigned by operators ("unjustified", last 3 months, up to 100). Use to find pending justification work or data-quality gaps per machine. Read-only: it cannot assign reasons.
list_devices
ChatGPTLists the IoT devices (sensors/telemetry gateways) the user can access, with their signal names and the machines they are attached to. Call before get_device_telemetry.
list_machines
ChatGPTLists the machines the user can access, with ids, names, machine group (production line/cell) and plant. Call this first: every other tool identifies machines by id. Use "search" or "groupId" to find a specific line or machine.
list_production_orders
ChatGPTProduction orders (OPs / job codes) that ran on the selected machines during a period, with first and last time seen. Use to find an order code before calling get_production_order_metrics.
list_shifts
ChatGPTLists the shifts (work periods) configured for the user's machines. History tools aggregate data per shift; use a shift id when the user asks about a specific shift (e.g. "night shift"). A shift whose start equals its end covers the full day.
list_stop_reasons
ChatGPTLists the stop reasons (stop types) configured in Hyperus, with group, whether they affect availability, maintenance category and optional sub-reasons. Use it to translate a reason name the user mentions into an id for other tools.
search_products
ChatGPTSearches the product catalog by code or description and returns ids, codes, descriptions and unit factors (weight, length, density). Use to identify a product the user mentions.