Sheets

Import data, manage sheet tabs, add rows, and scope Table work by sheet.

A sheet is a tab inside a Table. Each sheet has its own rows, columns, scoring, version history, and analytics.

Use multiple sheets when related workflows belong together but need different inputs, columns, or test cases. For example, keep a main regression sheet, edge cases, and migration work in the same Table.

Sheet anatomy

A sheet contains:

Part	Purpose
Rows	Examples, request logs, test cases, or batch items.
Text columns	Editable inputs, labels, expected outputs, metadata, and notes.
Computed columns	Prompt, code, assertion, extraction, comparison, composition, or helper work.
Score configuration	The scoring rules for the current sheet.
History	Saved versions and score history for the sheet.
Analytics	Request-level analytics scoped to the current Table and sheet.

Add rows manually

Each row is one item to run through the sheet. Use the row control at the bottom of the grid to add rows, then fill the input text columns.

Manual rows are useful for small test sets, edge cases, and examples you want to type or paste directly into the sheet.

Upload data

Use Upload when your rows already exist outside PromptLayer. Upload a CSV from your computer or import logged requests from request history.

Create a blank sheet via API

When building a table programmatically, you can create an empty sheet without importing data first. If you do not want a default Sheet 1 when creating the table, pass create_default_sheet: false to Create Table, then call POST /api/public/v2/tables/{table_id}/sheets with an optional title and omit source. The response includes a sheet with a default text column and one empty row. Then add columns and rows incrementally.

This is useful when you want to define the column structure first, then populate rows, without seeding a throwaway file.

After import, add computed columns and scoring in the same sheet.

Import request history

Choose Upload then From request history to open the request import dialog.

Each selected request becomes a row. Use request-history imports to evaluate real traffic, build datasets from production behavior, or replay previous requests through new columns.

Download a sheet

Click Download in the toolbar to export the current sheet for external review, offline analysis, or a point-in-time snapshot.

Use multiple sheets

Use the sheet tabs at the bottom of the Table to move between sheets. Each sheet can have different rows, columns, scoring, and history.

Use separate sheets for:

• Main regression set versus edge cases.
• Different task families in the same product area.
• Prompt variants that need different inputs.
• Migration work where the old and new workflows should stay near each other.
• Intermediate sheets that feed composition columns.

API references

Create sheet

Add a sheet to a Table.

List sheets

List the sheets in a Table.

Add rows

Add rows to a sheet programmatically.

Import request logs

Add logged requests to a sheet.