Skip to main content
Loading navigation
Back to prompt library

Build a Bilingual Facilities Maintenance Evidence Dossier Generator — implementation

Linked challenge: Build a Bilingual Facilities Maintenance Evidence Dossier Generator

Prompt source

implementation73 lines11 sectionsNo variables1 code block
## Build
Extract cooling plant work logs, sensor anomalies, and contractor closeout notes in Arabic and English. Synthesize verified bilingual maintenance packets with structured asset condition summaries for regional facility supervisors.

## User
Regional Facilities Operations Manager

## Workflow
1. Ingest bilingual field maintenance logs, mechanical inspection tags, and chiller differential pressure telemetry.
2. Align Arabic technical descriptions with standardized English maintenance defect classifications and asset tag numbers.
3. Check operational chiller parameters against EnergyPlus baseline chilled water supply cooling curves.
4. Synthesize a bilingual maintenance audit packet detailing completed corrective work and pending mechanical verifications.

## Interface
Dual-pane web console displaying parallel Arabic and English work records, condition rubrics, and a one-click export button for PDF dossiers.

## Data
Use mock normalized JSON logs matching Arabic and English field tickets, paired with synthetic EnergyPlus chiller curve references from epJSON test fixtures.
Starter examples below are synthetic. They are not downloaded records.

## Source
[EnergyPlus](https://raw.githubusercontent.com/NREL/EnergyPlus/develop/README.md)
References EnergyPlus documentation on epJSON input schema and baseline HVAC chiller performance curve objects to evaluate measured operational parameters.

## Tools
pydantic-ai, fastapi, weasyprint.
Pydantic AI provides strict type validation and structured output guarantees for multilingual ticket reconciliation and audit schemas.
Requires strong Arabic and English cross-lingual technical comprehension via hosted Claude API.

## 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
- Emits a normalized bilingual maintenance evidence dossier containing parallel Arabic and English fields.
- Accurately flags chiller differential pressures exceeding normal baseline curve parameters by over 15%.
- Generates deterministic sign-off status flags without silent record omission.

## Starter Examples
```json
[
  {
    "input": {
      "ticket_id": "DXB-CH-401",
      "asset_id": "CHILLER-02",
      "lang_ar": "تم تبديل صمام السحب وفحص ضغط التبريد ووجد تسريب بسيط",
      "lang_en": "Replaced suction valve and inspected refrigerant pressure, noted minor leak",
      "operating_delta_p_kpa": 142.5,
      "baseline_rated_delta_p_kpa": 120
    },
    "expected_output": {
      "dossier_id": "DOSSIER-DXB-CH-401",
      "asset_tag": "CHILLER-02",
      "summary_ar": "تم استبدال صمام السحب مع وجود تسريب ثانوي في سائل التبريد يتطلب متابعة",
      "summary_en": "Suction valve replaced; secondary refrigerant leak requires follow-up inspection",
      "delta_p_deviation_pct": 18.75,
      "exceeds_threshold": true,
      "verification_status": "PENDING_SUPERVISOR_SIGN_OFF"
    }
  },
  {
    "input": {
      "ticket_id": "DXB-CH-402",
      "asset_id": "CHILLER-05",
      "lang_ar": "فحص دوري للمضخة ولا توجد ملاحظات",
      "lang_en": "",
      "operating_delta_p_kpa": 118,
      "baseline_rated_delta_p_kpa": 120
    },
    "expected_output": {
      "dossier_id": "DOSSIER-DXB-CH-402",
      "asset_tag": "CHILLER-05",
      "summary_ar": "فحص دوري ناجح بدون ملاحظات",
      "summary_en": "Routine inspection successful with no recorded anomalies",
      "delta_p_deviation_pct": -1.67,
      "exceeds_threshold": false,
      "verification_status": "APPROVED"
    }
  }
]
```

## Safety
Maintenance dossiers are decision-support summaries only; physical valve manipulations and plant sign-offs require certified on-site mechanical engineer authorization.
Use mock external writes. Require human review before any real deployment.