Inventory / YAML

Apply written rules to a synthetic stock list containing a negative quantity and a ceramic mug assigned to stationery.

API key required

Set export TYPESAFE_API_KEY="your-api-key" with a key from the TypeSafe dashboard. Full setup instructions.

Get the files

Download items.yaml and jev_config.py into the same directory, or use examples/inventory/ in a checkout.

The config contains category rules and a selected correct example. Use it entirely through the CLI:

jevotron preview items.yaml --config jev_config.py
jevotron scan items.yaml --config jev_config.py

Quantity, category, and unit price receive separate scores. The entire item, including its description, remains visible. Review the cable quantity and mug category; exact model scores may vary.

Actual output

Captured from the command above. This run reused saved Jev assessments; the original assessment dates and probabilities are preserved. The tables render actual JSONL results, and future model results may differ.

Model: jev-1.13.0 · Assessment date: 2026-09-24.

Terminal summary (stderr):

Assessed 3 entries (3 cached); 2 warnings; emitted 3 entries.

All entries:

Entry	Label	Score	Warning
notebook	NORMAL	0.00	No
cable	ANOMALY	1.00	Yes
mug	ANOMALY	0.98	Yes

Flagged fields (score ≥ 0.5):

Entry	Field	Value	Score
cable	/quantity	-30	1.00
mug	/category	"stationery"	0.98

One complete warning entry (actual JSON)

{
  "assessed_at": "2026-09-24T00:23:46.803830+00:00",
  "cached": true,
  "fields": [
    {
      "confidence": 1.0,
      "label": "NORMAL",
      "path": "/category",
      "probabilities": {
        "ANOMALY": 0.0,
        "NORMAL": 1.0
      },
      "score": 0.0,
      "value": "electronics"
    },
    {
      "confidence": 1.0,
      "label": "ANOMALY",
      "path": "/quantity",
      "probabilities": {
        "ANOMALY": 1.0,
        "NORMAL": 0.0
      },
      "score": 1.0,
      "value": -30
    },
    {
      "confidence": 0.99,
      "label": "NORMAL",
      "path": "/unit_price_usd",
      "probabilities": {
        "ANOMALY": 0.0,
        "NORMAL": 1.0
      },
      "score": 0.0,
      "value": 8.0
    }
  ],
  "id": "cable",
  "label": "ANOMALY",
  "model": "jev-1.13.0",
  "request_hash": "4e0d609f167eb3e358d1a0f5e7cdc9326911c22889e6cfaf3aba017fcd5c1171",
  "score": 1.0,
  "source": "items.yaml:document:1:entry:2",
  "usage": {
    "input_tokens": 989,
    "output_tokens": 105
  },
  "warning": true
}

Download complete JSONL output · Download terminal summary

Download the actual warning CSV, or create it yourself:

jevotron scan items.yaml --config jev_config.py --warnings-only --output-format csv -o warnings.csv

Use flags instead

To check only quantity with a one-line instruction:

jevotron scan items.yaml --id-column id --field /quantity \
  --guidance "Quantity is physical stock and must be nonnegative." --warnings-only

This is a different assessment request and creates separate cache records.