Dataset Viewer
Auto-converted to Parquet Duplicate
case_id
stringclasses
3 values
category
stringclasses
3 values
severity
stringclasses
2 values
synthetic
bool
1 class
baseline_schema
dict
candidate_schema
dict
expected_result
stringclasses
2 values
rationale
stringclasses
3 values
synthetic-required-field-removed
missing_required_field
high
true
{ "type": "object", "required": [ "account_id", "email", "phone_number" ], "properties": { "account_id": { "type": "string" }, "email": { "type": "string" }, "phone_number": { "type": "string" } } }
{ "type": "object", "required": [ "account_id", "email" ], "properties": { "account_id": { "type": "string" }, "email": { "type": "string" } } }
block
A required field present in the baseline is absent from the candidate contract.
synthetic-field-type-changed
field_type_changed
high
true
{ "type": "object", "required": [ "order_id", "total_cents" ], "properties": { "order_id": { "type": "string" }, "total_cents": { "type": "integer" } } }
{ "type": "object", "required": [ "order_id", "total_cents" ], "properties": { "order_id": { "type": "string" }, "total_cents": { "type": "string" } } }
block
The field name is stable but its declared type changes from integer to string.
synthetic-nested-field-moved
nested_field_moved
medium
true
{ "type": "object", "required": [ "sku", "warehouse" ], "properties": { "sku": { "type": "string" }, "warehouse": { "type": "object", "required": [ "region" ], "properties": { "region": { "type": "string" } } } } }
{ "type": "object", "required": [ "sku", "fulfilment" ], "properties": { "sku": { "type": "string" }, "fulfilment": { "type": "object", "required": [ "warehouse_region" ], "properties": { "warehouse_region": { "type": "string" }...
review
The same conceptual value appears to have moved to a different object path; human mapping review is required.

Synthetic Schema Drift Regression Preview

Three deliberately small, synthetic cases for testing how a schema-change review or regression gate handles common contract changes:

  • a removed required field;
  • a field type change; and
  • a moved nested field.

Every record is fictional and marked synthetic: true. This preview contains no customer data, credentials, production API response, web extraction, or paid-product file.

Contents

schema-drift-preview.jsonl contains one JSON object per case. Each object has a baseline_schema, candidate_schema, expected_result, and a concise rationale. It is designed for local test fixtures, examples, and evaluation scaffolding—not for production compatibility decisions by itself.

Limitations

  • The three cases are illustrative, not a complete schema-drift taxonomy.
  • No real API, dataset, or service is represented.
  • Consumers remain responsible for their own compatibility policy, test coverage, deployment checks, and source-right review.
  • This is not a hosted API, monitoring service, security assurance, or guarantee that all breaking changes will be detected.

Related paid developer tool

For a self-hosted Node.js tool that compares bounded dataset snapshots for row, field, type, and schema drift, see the Dataset Diff and Schema Drift API.

Release checklist

This local package is prepared for a future Hugging Face dataset repository. Before publishing: verify the NimbliqueStudio organization, choose a public repository name, upload these exact files without modification, confirm the generated public URL while signed out, record that URL in nimblique-marketing-channels.md, and then log the first promotion in publication-log.csv.

Downloads last month
36