Recipe catalog / qualification-evidence

Check evidence for a qualification

Does profile contain concrete evidence that the candidate meets requirement, not just matching keywords?

You screen resumes or candidate summaries against one requirement at a time and want to separate demonstrated experience from keyword matches.

Explore this recipe interactively · Source and implementation guide

Use qualification-evidence 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 { qualificationEvidence } from 'jev-recipes/qualification-evidence';

const result = await qualificationEvidence({
  "requirement": "At least three years of experience operating Kubernetes clusters in production.",
  "profile": "Platform Engineer, Northwind Logistics, 2020 to present. Own the production Kubernetes platform (EKS, 12 clusters, 400 services). Led the migration from self-managed clusters to EKS in 2022 with zero customer-facing downtime. On-call rotation for cluster incidents; wrote the runbooks for node pool upgrades and etcd recovery. Skills: Kubernetes, Terraform, Go, Prometheus.",
  "minConfidence": 0.8
});
console.log(result);

Input contract

{
  "input": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "requirement": {
        "type": "string"
      },
      "profile": {
        "type": "string"
      },
      "minConfidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      }
    },
    "required": [
      "requirement",
      "profile"
    ]
  },
  "result": {
    "$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": [
          "evidenced",
          "unevidenced"
        ]
      }
    },
    "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 qualification-evidence.

{
  "model": "demo-fixture",
  "usage": {
    "input_tokens": 0,
    "output_tokens": 0
  },
  "status": "ready",
  "probability": 0.94,
  "confidence": 0.94,
  "verdict": "evidenced"
}

Evaluation evidence

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

Limitations

• Judges whether the profile describes evidence, not whether that evidence is true or verifiable.
• One requirement per call. Loop over requirements or aggregate in application code.
• A profile can be evidenced for a requirement and still be a poor fit overall; this is not a hiring decision.

Related recipes

• step-complete: Use step-complete for the general form: does evidence establish that a condition is met.
• verify: Use verify to check several candidate claims against paired evidence in one call.