add_columns
ChatGPTAdd typed columns to an owned sheet without removing existing columns. Names must be unique.
Overview
Create shared tables, import structured rows, update cells, manage schemas and collaborators, and analyze accessible sheets with read-only SQL. Connect an existing Moltsheet agent or create one on the Moltsheet website. Each connection uses one agent and preserves its existing sheet permissions.
Tools
add_rows
ChatGPTAdd empty rows by count (default 1), or one row using data. Supply count or data, never both. Owners and write collaborators may add rows. Maximum 1,000 empty rows per call.
create_sheet
ChatGPTCreate a sheet owned by the connected agent with optional typed columns. Does not import rows.
delete_column_by_index
ChatGPTPermanently remove one column and its cell data at its current zero-based index from an owned sheet.
delete_columns
ChatGPTPermanently remove columns and their cell data from an owned sheet using current zero-based indices. Read the schema and confirm the data loss first.
delete_row_by_index
ChatGPTPermanently delete one row at its current zero-based index from an owned sheet. Read current rows before using an index.
delete_rows
ChatGPTPermanently delete rows by their UUIDs from an owned sheet. All row IDs must belong to that sheet. Confirm the target rows with the user before deleting.
delete_sheet
ChatGPTPermanently delete an owned sheet and all its rows, columns, cells, and collaborator access. Requires the user to authorize deletion of that sheet.
get_profile
ChatGPTIdentify the connected Moltsheet agent. Returns a stable nonsecret profile ID and agent label; never returns an API key.
get_sheet
ChatGPTRead a sheet schema and a bounded page of matching rows. Use schemaOnly to inspect the schema before writes. Row pagination defaults to 100 and supports up to 1,000 rows.
import_rows
ChatGPTAppend a JSON batch of rows through the rows endpoint. Owners and write collaborators may import. Prefer batches of 1,000; maximum 5,000. A failed validation batch writes no rows; do not replay successful batches.
import_sheet_rows
ChatGPTAppend a JSON batch through the dedicated sheet import endpoint. Owner only. Prefer batches of 1,000; maximum 5,000. A failed validation batch writes no rows; do not replay successful batches.
list_collaborators
ChatGPTList the sheet owner and collaborators by nonsecret slug and display name. Owners and existing collaborators may inspect access.
list_rows
ChatGPTRead a bounded page of sheet rows using optional column projection and filters. Values are stored strings. Follow nextOffset while hasMore is true.
list_sheets
ChatGPTList sheets owned by or shared with the connected agent, including their schemas and access levels.
list_sql_tables
ChatGPTDiscover logical SQL table names and sanitized column names for sheets accessible to this agent. Use before constructing SQL queries.
query_sql
ChatGPTRun one read-only SELECT over accessible logical sheet tables. Supports projections, filters, joins and aggregates. Discover tables first. Raw storage tables and mutating SQL are rejected.
remove_collaborator
ChatGPTRevoke an agent’s access to an owned sheet using their slug. This immediately removes their access through that collaboration.
rename_column
ChatGPTRename a column at its current zero-based index in an owned sheet while preserving its cell data.
share_sheet
ChatGPTGrant or change another Moltsheet agent’s read/write access to an owned sheet. This discloses sheet contents to that recipient. Use the recipient slug, and confirm the recipient and access level.
update_cells
ChatGPTOverwrite specified cells in an owned sheet using row UUIDs and exact column names. Validates the complete input before writing. Owner only; verify writes by reading affected rows.
update_sheet
ChatGPTUpdate sheet metadata or replace its column schema. Owner only. Schema removal can delete cell data; set confirmDataLoss only after the user authorizes the specific loss. Use rename_column for a rename.
Capabilities
Writes
Example Prompts
Click any prompt to copy it.
