Build a Sparkplug B Nuisance Alarm Suppressor and Aggregator — implementation

## Build
Correlate repetitive Sparkplug B DDATA alarm payloads to identify oscillating sensor nuisance trips. Synthesize an aggregated alarm dashboard that groups related warnings and suppresses redundant notifications for operators.

## User
Control Room Operator

## Workflow
1. Ingest continuous Sparkplug B DDATA streams containing raw alarm state transitions and metric timestamps.
2. Identify rapidly toggling sensor states that cross threshold limits multiple times within a short observation window.
3. Group related nuisance alarms originating from the same equipment node and calculate a composite severity score based on the toggle count.
4. Output an aggregated incident ticket that suppresses individual toggle noise while preserving the core equipment fault for the dashboard.

## Interface
Real-time alerting dashboard displaying aggregated equipment incidents, suppressed alarm toggle counts, and historical Sparkplug B metric timelines.

## Data
Generate a continuous stream of Sparkplug B JSON payloads featuring nominal states mixed with high-frequency oscillating alarm states across multiple equipment nodes.
Starter examples below are synthetic. They are not downloaded records.

## Source
[Eclipse Sparkplug](https://raw.githubusercontent.com/eclipse-sparkplug/sparkplug/master/README.md)
Uses Sparkplug B DDATA specification structures to represent realistic stateful device metric updates and oscillating sensor transitions.

## Tools
mastra-ai, fastapi, react.
Mastra AI provides stateful agent memory and data aggregation patterns to analyze time-windowed sensor events and group nuisance alarms.
Requires structured data parsing and temporal reasoning capabilities to group related state transitions using local or hosted LLMs.

## Deliverables
Submit runnable source, pinned dependencies, setup commands, a working interface, example outputs, and a recorded end-to-end demo. Include tests and an action log.

## Acceptance
- Successfully detects alarms toggling between normal and fault states within the defined suppression window.
- Groups oscillating events into a single aggregated incident ticket with an accurate toggle count.
- Maintains the correct original timestamp for the first instance of the fault while suppressing subsequent redundant notifications.

## Starter Examples
```json
[
  {
    "input": {
      "node_id": "Line1/Feeder3",
      "alarms": [
        {
          "timestamp": "2025-06-01T14:09:40Z",
          "metric": "Vibration",
          "state": "HIGH"
        },
        {
          "timestamp": "2025-06-01T14:09:42Z",
          "metric": "Vibration",
          "state": "NORMAL"
        },
        {
          "timestamp": "2025-06-01T14:09:45Z",
          "metric": "Vibration",
          "state": "HIGH"
        }
      ],
      "suppression_window_sec": 10,
      "schema_note": "Simplified synthetic app records, not complete official schema files"
    },
    "expected_output": {
      "node_id": "Line1/Feeder3",
      "nuisance_detected": true,
      "toggle_count": 2,
      "aggregated_incident": {
        "metric": "Vibration",
        "status": "OSCILLATING_HIGH",
        "first_seen": "2025-06-01T14:09:40Z"
      },
      "suppressed_events": 2
    }
  },
  {
    "input": {
      "node_id": "Line1/Chiller",
      "alarms": [
        {
          "timestamp": "2025-06-01T21:15:00Z",
          "metric": "Temp",
          "state": "HIGH"
        }
      ],
      "suppression_window_sec": 10,
      "schema_note": "Simplified synthetic app records, not complete official schema files"
    },
    "expected_output": {
      "node_id": "Line1/Chiller",
      "nuisance_detected": false,
      "toggle_count": 0,
      "aggregated_incident": {
        "metric": "Temp",
        "status": "HIGH",
        "first_seen": "2025-06-01T21:15:00Z"
      },
      "suppressed_events": 0
    }
  }
]
```

## Safety
Outputs advisory alarm aggregations for operator review; does not actively modify machine configurations or bypass physical safety interlocks.
Use mock external writes. Require human review before any real deployment.