Recipe catalog / maintenance-hazard-wording

Flag a maintenance request that describes a safety hazard

Does request describe a safety hazard such as a gas smell, active water intrusion, exposed wiring, structural damage, no heat in cold weather, or a blocked exit?

You need to pull tenant maintenance requests that describe a hazard out of the routine queue for immediate dispatch, based on what the request says before anyone inspects.

Explore this recipe interactively · Source and implementation guide

Use maintenance-hazard-wording in TypeScript

Install with npm install jev-recipes. Requires Node.js 22.9 or newer and ES modules. Set TYPESAFE_API_KEY in your server environment for live calls, which send input to TypeSafe and use API quota. See the installation guide.

import { maintenanceHazardWording } from 'jev-recipes/maintenance-hazard-wording';

const result = await maintenanceHazardWording({
  "request": "Unit 4B. Since this morning there's a strong gas smell in the kitchen, strongest near the stove even with all the burners off. I opened the windows. Can someone come take a look when they get a chance? Thanks.",
  "minConfidence": 0.8
});
console.log(result);

Input contract

{
  "input": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "request": {
        "type": "string"
      },
      "minConfidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      }
    },
    "required": [
      "request"
    ]
  },
  "result": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "model": {
        "type": "string"
      },
      "usage": {
        "type": "object",
        "properties": {
          "input_tokens": {
            "type": "integer",
            "minimum": 0,
            "maximum": 9007199254740991
          },
          "output_tokens": {
            "type": "integer",
            "minimum": 0,
            "maximum": 9007199254740991
          }
        },
        "required": [
          "input_tokens",
          "output_tokens"
        ],
        "additionalProperties": false
      },
      "status": {
        "type": "string",
        "enum": [
          "ready",
          "review"
        ]
      },
      "probability": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "confidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "verdict": {
        "type": "string",
        "enum": [
          "hazard",
          "routine"
        ]
      }
    },
    "required": [
      "model",
      "usage",
      "status",
      "probability",
      "confidence",
      "verdict"
    ],
    "additionalProperties": false
  }
}

Saved example result

This hand-authored response demonstrates the contract. It is not a model accuracy measurement. Run it without an API key: npx jev-recipes demo maintenance-hazard-wording.

{
  "model": "demo-fixture",
  "usage": {
    "input_tokens": 0,
    "output_tokens": 0
  },
  "status": "ready",
  "probability": 0.96,
  "confidence": 0.96,
  "verdict": "hazard"
}

Evaluation evidence

Use the evaluation guide to measure this decision on your own labeled cases.

Limitations

• Judges the wording of the request, not the actual condition of the property. Only an inspection establishes whether a hazard exists.
• Applies a fixed set of hazard types drawn from common habitability standards; local codes and lease obligations belong in application rules.
• Does not rank hazards against each other or decide dispatch timing.

Related recipes

• urgency-signal: Use urgency-signal to detect whether the tenant explicitly asks for urgent attention, rather than whether the described problem is a hazard regardless of how it is asked.
• issue-impact: Use issue-impact to grade how badly a reported problem affects the reporter, rather than whether it is a safety hazard.