Example Prompts
This page shows real prompts you can use with an AI coding assistant connected to MaestroHub via MCP, along with the actual tool calls and outputs they produce.
Exploring Connections
"What connections are available for the Caliperline system?"
The AI calls connectors_list_connections with search "Caliperline" and returns:
Found 8 Caliperline connections:
| Name | Type | Description |
|-------------------------|----------|--------------------------|
| Caliperline-OP10-CNC | Modbus | Turning station |
| Caliperline-OP20-CNC | OPC UA | CNC finish machining |
| Caliperline-OP30-WASH | Modbus | Washing station |
| Caliperline-OP40-TEST | OPC UA | Testing station |
| Caliperline-MES-API | REST | Manufacturing Execution |
| Caliperline-ERP-API | REST | Enterprise Resource Plan |
| Caliperline-MQTT-Broker | MQTT | Message broker |
| Caliperline-PostgreSQL | Postgres | Factory database |
Reading Live Telemetry
"Show me the latest telemetry from the OP10 CNC machine"
The AI calls uns_fetch_recent on topic mHv1.0/caliperline/machines/op10/telemetry and returns:
{
"machine": "OP10-CNC",
"State": 3,
"SpindleSpeed": 500,
"SpindleLoad": 62,
"Temperature": 262,
"Vibration": 129,
"PowerKW": 358,
"PartsProduced": 32,
"PartsGood": 31,
"PartsBad": 1,
"FaultActive": 1,
"FaultCode": 4,
"RunTime": 67436544,
"DownTime": 0
}
The AI can then interpret this data: "OP10 is in fault state (State=3) with fault code 4 active. Spindle is at 500 RPM / 62% load. Temperature is 262°C. 32 parts produced with a 96.9% quality rate (31 good, 1 bad)."
Searching Topics
"What UNS topics exist for Caliperline?"
The AI calls uns_search_topics with query "caliperline" and returns:
Found 28 topics under the Caliperline namespace:
mHv1.0/caliperline/
├── erp/
│ └── inventory
├── inventory/
│ └── status
├── machines/
│ └── op10/
│ └── telemetry ← live Modbus data every 10s
├── maintenance/
│ └── status
├── mes/
│ └── orders/
│ └── active ← synced from MES every 1m
└── alerts/
└── op20/
└── fault ← fault alerts from OP20
Inspecting Pipelines
"What pipelines are running for Caliperline?"
The AI calls engine_list_pipelines with search "Caliperline" and returns:
Found 3 Caliperline pipelines:
1. Caliperline OP10 Telemetry (enabled)
Polls Modbus telemetry from OP10 every 10s → publishes to UNS
Nodes: Schedule Trigger → Modbus Read Group → JavaScript → UNS Publish
2. Caliperline OP20 Fault Alerts (enabled)
Monitors OP20 OPC UA for faults every 5s → publishes alerts
Nodes: Schedule Trigger → OPC UA Read Group → Condition → JavaScript → UNS Publish
3. Caliperline MES Orders Sync (enabled)
Fetches active orders from MES REST API every 1m → publishes to UNS
Nodes: Schedule Trigger → REST Request → JavaScript → UNS Publish
Creating a Dashboard
"Create a dashboard to monitor OP10 machine health"
The AI first calls uns_fetch_recent to understand the data schema and value ranges, then calls uns_create_dashboard. Here's what it produces:
Dashboard "Caliperline OP10 Machine Health" created with 9 panels:
Row 1: Machine State (state-timeline, full width)
- Shows Running/Idle/Fault states over time
Row 2: Gauges
- Spindle Speed (0–10,000 RPM) with warning at 8,000+
- Spindle Load (0–100%) with warning at 75%+
- Vibration trend line with threshold bands at 70 and 90 mm/s
Row 3: More gauges + quality
- Temperature (0–500°C) with warning at 300+
- Power Consumed (0–1,000 kW) with warning at 800+
- Production Quality donut chart (Good vs Bad parts)
Row 4: MES Orders table (all columns, sorted by start time)
Row 5: OP20 Fault Alerts table (timestamp, code, name, severity)
Dependency Analysis
"What would be affected if I delete the Caliperline OP20 connection?"
The AI calls deps_query_impact for the OP20 connection and returns:
Impact analysis for Caliperline-OP20-CNC:
Directly affected:
- 3 OPC UA functions (Read FaultActive, Read FaultCode, Read FaultName)
Indirectly affected:
- Pipeline "Caliperline OP20 Fault Alerts" — uses all 3 functions via
OPC UA Read Group node
- UNS topic "mHv1.0/caliperline/alerts/op20/fault" — would stop receiving
new fault alerts
⚠️ Deleting this connection will break 1 pipeline and stop fault
monitoring for OP20.
Finding Unused Resources
"Are there any unused connections I could clean up?"
The AI calls deps_query_orphans filtered to connections and returns:
Found 5 connections with no pipeline dependencies:
| Connection | Type | Created |
|-------------------------|----------|------------|
| Caliperline-PostgreSQL | Postgres | 2026-03-10 |
| Caliperline-OP30-WASH | Modbus | 2026-03-10 |
| Caliperline-OP40-TEST | OPC UA | 2026-03-10 |
| Caliperline-ERP-API | REST | 2026-03-10 |
| MSTeams | — | — |
These connections exist but aren't used by any pipeline. You may want
to create pipelines for them or remove them to reduce clutter.
Building a Pipeline
"Create a pipeline that reads OP10 telemetry every 30 seconds and alerts when vibration exceeds 100 mm/s"
The AI calls engine_list_node_types and connectors_list_functions to discover available nodes, then calls engine_create_pipeline:
Pipeline "OP10 Vibration Alert" created and enabled.
Flow:
Schedule Trigger (30s)
→ Modbus Read Group (OP10, all registers)
→ Condition (Vibration > 100)
→ [true] JavaScript (format alert payload)
→ UNS Publish → mHv1.0/caliperline/alerts/op10/vibration
The pipeline will check OP10 vibration every 30 seconds. When it
exceeds 100 mm/s, an alert is published to the UNS with machine ID,
current vibration value, timestamp, and severity level.
Tips for Effective Prompts
- Be specific about data — mention topic paths, connection names, or field names when you know them
- Ask for exploration first — prompts like "What topics exist under caliperline?" help you discover what's available before building on it
- Iterate — start with "Create a simple pipeline..." then follow up with "Add a condition node that..."
- Ask for impact before changing — "What depends on this connection?" before deleting anything
- Reference existing resources — "Use the same connection as the OP10 Telemetry pipeline" helps the AI reuse your setup