Skip to content

Create and Edit Datasets

Use New Dataset to open a two-step wizard: metadata first, then table data.

Dataset wizard metadata step

Metadata Step

Field Required Description
Name Yes Human-readable dataset name.
Type Yes Manual, Pytest Parametrize, or Imported.
Description No What the data covers and how it should be used.
Source No Reference to file, test, story, or external system.

Import CSV/JSON

The metadata step can import CSV or JSON. Importing:

  • reads headers as future columns;
  • creates rows from file records;
  • suggests name and source ref from the filename when empty;
  • changes source type to Imported;
  • reports imported row and column counts.

Use import when data already exists in a spreadsheet, automation fixture, BI export, or production-like source.

Table Builder

The Table Builder is a spreadsheet-like editor. Users can:

  • edit headers by clicking column names;
  • edit cells inline;
  • add columns from the header;
  • delete columns;
  • add rows below the current row;
  • delete rows;
  • move through cells with Enter or Tab;
  • cancel inline editing with Escape;
  • work with large tables through virtual scrolling.

Validation

The editor enforces:

  • at least one column;
  • at most 10 columns;
  • at least one row;
  • non-empty headers;
  • unique normalized headers;
  • unique non-empty row keys.

Dataset table builder

Columns

Each API column has:

  • column_key;
  • display_name;
  • data_type;
  • required;
  • default_value;
  • is_scenario_label.

The UI generates column_key from the header. Renaming a header moves row values to the new key.

Rows

Each API row has:

  • row_key;
  • scenario_label;
  • values;
  • is_active.

The UI shows cell values. Detailed row metadata is available through the API payload.

Dataset Details

Opening a dataset shows name, id, source type, status, linked cases, current revision, column and row counts, source ref, updated timestamp, description, and current revision table.

Edit Dataset

Editing uses the same wizard with existing values. Saving creates a new revision when structure or rows changed.

Delete and Bulk Delete

Deleting removes the dataset from the project and removes all test case bindings. Prefer revisions over deletion when historical context matters.