Connect AI Googlesheets

CData Connect AI — Google Sheets Skill

⚠️ Prerequisites — load these first

• connect-ai-base skill

Before proceeding, locate and read the connect-ai-base skill. If it is not available in the current environment (not loaded, not enabled, or not installed), stop immediately. Do not proceed with the task. Tell the user that the connect-ai-base skill is required and ask them to install and enable it before continuing.

This skill provides Google Sheets-specific guidance for querying Google Sheets data through CData Connect AI. It composes on top of the connect-ai-base skill, which handles the discovery workflow, error recovery, SQL dialect, and query naming convention.

Precedence

This skill replaces getInstructions for the Google Sheets driver. Do not call getInstructions for Google Sheets — the guidance it provides is already incorporated here. Proceed directly to schema discovery (getTables / getColumns) after identifying the Google Sheets connection via getCatalogs.

Schema

Google Sheets is a single-schema driver. The schema name is GoogleSheets.

Use the three-part name [Catalog].[Schema].[Table], where the catalog is the connection name:

SELECT * FROM [YourConnection].[GoogleSheets].[Spreadsheets]

Query Process

Google Sheets data is organized as spreadsheets → sheets → cell data. Work down that hierarchy:

1. Identify the spreadsheet. If you know the spreadsheet or sheet name, call getTables with a tableName filter (e.g. %budget%) to find the matching data tables directly. If you don't, query the Spreadsheets table first, filtering on ModifiedTime or OwnerEmail to narrow the list rather than returning everything.
2. Explore the sheets within it. Query the Sheets table filtered by SpreadsheetId to see the worksheets, their dimensions, and their names.
3. Query the sheet's data. Each worksheet is exposed as its own data table named [SpreadsheetName_SheetName]. Run getColumns on that table first — its columns come from the sheet's header row and are not knowable in advance — then query the data.

Data Model

Key Tables

• Spreadsheets — metadata for every Google Sheets document the connected user can access (one row per spreadsheet). Read-only.
• Sheets — one row per worksheet (tab) across all accessible spreadsheets, with dimensions and properties. Read-only.
• Folders — the Google Drive folder structure containing the spreadsheets.
• [SpreadsheetName_SheetName] — dynamic data tables. getTables returns one of these for every worksheet in every accessible spreadsheet, presented as a flat list alongside the metadata tables. These hold the actual cell data and are where reads and writes of sheet content happen.

Key Relationships

• Sheets.SpreadsheetId → Spreadsheets.Id — each sheet belongs to one spreadsheet.
• Spreadsheets.ParentIds → Folders.Id — a spreadsheet's parent folders (comma-separated list).
• A [SpreadsheetName_SheetName] data table corresponds to one row in Sheets (its SpreadsheetName + SheetName).

Important Columns

Spreadsheets

• Id — unique spreadsheet identifier (primary key); used as the Id argument to the spreadsheet stored procedures.
• Name — display name of the spreadsheet.
• Description — short description.
• OwnerName, OwnerEmail — the owner's name and email.
• ModifiedTime, CreatedTime — last-modified and creation timestamps (TIMESTAMP); use these for time-based filtering.
• Trashed — whether the spreadsheet is in the trash (BOOLEAN).
• Starred, Viewed — whether the user has starred / viewed it (BOOLEAN).
• ParentIds — comma-separated list of parent folder Ids.

Sheets

• SpreadsheetId, SpreadsheetName — the parent spreadsheet's Id and name.
• SheetId — identifier for the worksheet within the spreadsheet.
• SheetName — display name of the worksheet.
• SheetIndex — zero-based position of the sheet within the spreadsheet.
• SheetType — sheet type; defaults to GRID.
• RowCount, ColumnCount — grid dimensions (INTEGER).
• FrozenRowCount, FrozenColumnCount — frozen rows/columns (INTEGER).
• Hidden — whether the sheet is hidden in the UI (BOOLEAN).

[SpreadsheetName_SheetName] data tables

• id — auto-generated INTEGER primary key. The driver adds this to every data table (even sheets that have no key in the source), so it is reliably available to key UPDATE on.
• Remaining columns come from the sheet's header row. Always run getColumns before querying — column names, spaces, and special characters vary per sheet.

Common Query Patterns

Recently modified spreadsheets

SELECT [Id], [Name], [ModifiedTime], [OwnerEmail]
FROM [YourConnection].[GoogleSheets].[Spreadsheets]
WHERE [ModifiedTime] >= '2026-01-01'
ORDER BY [ModifiedTime] DESC

Spreadsheets by owner

SELECT [Id], [Name], [OwnerName], [OwnerEmail]
FROM [YourConnection].[GoogleSheets].[Spreadsheets]
WHERE [OwnerEmail] = 'user@example.com'

Sheets within a specific spreadsheet

SELECT [SheetId], [SheetName], [RowCount], [ColumnCount]
FROM [YourConnection].[GoogleSheets].[Sheets]
WHERE [SpreadsheetId] = '<spreadsheet-id>'
ORDER BY [SheetIndex]

Data from a specific sheet

The data table is named [SpreadsheetName_SheetName]. Run getColumns first to confirm the column names (they come from the header row), then query:

SELECT *
FROM [YourConnection].[GoogleSheets].[Engineering OKR_Sheet1]
WHERE [Progress %] < 50
ORDER BY [Progress %] ASC

Stored Procedures

Call these with executeProcedure. Parameter names are passed with an @ prefix (e.g. @Title).

CreateSpreadsheet

Creates a new spreadsheet in the user's Google Drive. Returns the new spreadsheet Id.

{
  "catalogName": "YourConnection",
  "schemaName": "GoogleSheets",
  "procedureName": "CreateSpreadsheet",
  "parameters": { "@Title": "New Spreadsheet" }
}

Optional: @Description, @Starred, @Hidden, @Restricted, @Parents (parent folder Ids).

AddSheet

Adds a worksheet to an existing spreadsheet. Supply @HeaderNames so the new sheet has a header row you can insert against afterward. @SheetId must be a non-negative integer and cannot be changed once set.

{
  "catalogName": "YourConnection",
  "schemaName": "GoogleSheets",
  "procedureName": "AddSheet",
  "parameters": {
    "@SpreadsheetId": "<spreadsheet-id>",
    "@SheetId": "1",
    "@Title": "New Sheet",
    "@HeaderNames": "Header1,Header2,Header3"
  }
}

Enclose a header in double quotes inside @HeaderNames if it contains special characters. Optional: @Index, @RowCount, @ColumnCount, @FrozenRowCount, @FrozenColumnCount, @Hidden, @HideGridlines, @RightToLeft.

UpdateSheet

Updates properties of an existing worksheet/tab — @Title (the tab's name), dimensions, frozen rows/columns, visibility — identified by @SpreadsheetId and @SheetId (both required). @Title renames the tab, not the spreadsheet document. There is no procedure to rename a spreadsheet: the Spreadsheets table is read-only, so set a spreadsheet's name at creation via CreateSpreadsheet @Title, or rename it in the Google Sheets UI.

CopySheet

Copies a sheet from one spreadsheet into another.

DeleteSheet

Deletes a worksheet from a spreadsheet (by @SpreadsheetId and @SheetId).

DeleteSpreadsheet

Deletes an entire spreadsheet by its Id.

FormatRange

Applies cell formatting to a specified range within a sheet.

AddDataSource / AddDataSourceTable

Attach a BigQuery data source (and data-source-backed table) to a spreadsheet. These require a custom OAuth app with the bigquery.readonly scope and are only relevant for BigQuery-connected sheets.

DownloadDocument — Not currently supported in cloud

DownloadDocument returns the exported file via LocalFile (server-side disk path, unavailable in cloud) or FileStream (Java OutputStream, unable to pass through the MCP interface). It also exposes an Encoding (base64) retrieval path, but that path is not functional in the cloud either — do not attempt it as a workaround, and do not invent parameter names the driver does not accept. Treat document download as a current limitation: to read a spreadsheet's contents, query its data tables directly, or download it from the Google Sheets UI. Support for file downloads via stored procedures is planned — check for updates if this capability is needed.

UploadDocument — Not currently supported in cloud

UploadDocument accepts LocalFile (server-side disk path, unavailable in cloud) or Content (Java InputStream, unable to pass through the MCP interface). It also accepts a base64 string via FileData + Encoding, but that path is not functional in the cloud either — do not attempt it as a workaround, and do not invent parameter names the driver does not accept. Treat file uploads as a current limitation. To build a spreadsheet from scratch, use CreateSpreadsheet + AddSheet + INSERT (see Write Operations) instead. Support for file uploads via stored procedures is planned — check for updates if this capability is needed.

Write Operations

Write spreadsheet content by inserting and updating rows on the [SpreadsheetName_SheetName] data tables. Run getColumns first so you reference the sheet's real header columns.

INSERT INTO [YourConnection].[GoogleSheets].[Engineering OKR_Sheet1] ([Objective], [Owner], [Progress %])
VALUES ('Ship v2', 'A. Patel', '40')

-- Preferred: target the derived [id] for a deterministic single-row update
UPDATE [YourConnection].[GoogleSheets].[Engineering OKR_Sheet1]
SET [Progress %] = '75'
WHERE [id] = 4

Key UPDATE statements on the auto-generated [id] column — this matches the driver's documented WHERE Id = <expression> grammar and affects exactly one row. (DELETE FROM is not available over the MCP surface — no execute_delete; remove rows in the Sheets UI or drop the whole sheet with DeleteSheet.) The driver adds an id key to every data table, so it is always available. A WHERE clause on a non-key column is accepted but updates every row whose value matches (e.g. WHERE [Name] = 'Smoke Test' updates all rows named "Smoke Test"), so prefer [id] unless you intend a multi-row change. Parameterize values with @param (or ?) and pass them as strings for VARCHAR cells.

To build a new spreadsheet end to end: call CreateSpreadsheet, then AddSheet (with @HeaderNames to lay down the header row), then INSERT into the resulting [SpreadsheetName_SheetName] data table.

Write access is governed by the connection's mode in Connect AI. If inserts, updates, or write stored procedures are rejected, the connection is likely set to read-only — have the user enable write access in the Connect AI connection settings.

Google Sheets-Specific Conventions

• Data tables are dynamic and per-sheet. Every worksheet becomes its own table named [SpreadsheetName_SheetName]. getTables returns these as a flat list mixed in with the Spreadsheets, Sheets, and Folders metadata tables.
• Always getColumns a data table before querying it. Its columns derive from the sheet's header row, so they are unknown until inspected and differ from sheet to sheet.
• Special characters in names are preserved by default. Spreadsheet and sheet names keep their spaces and symbols in table names (e.g. [BI Application Testing_Power BI]), so always bracket them. They are only replaced with underscores if the connection sets UseSimpleNames=true, which is off by default.
• Bracket every identifier. Spreadsheet, sheet, and column names routinely contain spaces and symbols (e.g. [Progress %]); always wrap them in [ ].
• Column data types default to VARCHAR. Sheet cell columns are typically untyped text. Be deliberate about numeric and date comparisons, and pass values as strings in INSERT/UPDATE.
• Data tables have an auto-generated id primary key. The driver adds an INTEGER id key column to every data table, so a deterministic single-row key is reliably available for UPDATE.
• Preview before pulling full sheets. Use LIMIT to sample large sheets before running unbounded queries.
• Metadata tables are read-only. Spreadsheets and Sheets cannot be written to — manage spreadsheets and sheets through the stored procedures, and write cell data through the data tables.