NocoDB n8n integrationAutomate NocoDB with n8n.

Writes a new row in the selected table and returns it with the ID NocoDB assigned, ready for the next node.

• Base Name or ID: chosen from the list, or set with an expression, the small n8n formula that reads the incoming item.
• Table Name or ID: the table to operate on, listed once the base is selected.
• Fields to Send: the column mapper. With Data to Send on autoMapInputData, incoming properties fall into columns of the same name; mapWithFields lets you fill each column yourself, and Inputs to Ignore drops the properties you do not want stored.

One operation for both cases: the row exists and gets updated, or it does not and gets created. The duplicate problem disappears from the workflow logic.

• Row ID Value: the deciding field. Filled, NocoDB updates that row; left empty, it inserts a new one, which is exactly what {{ $json.id }} does when the previous node sometimes returns a match and sometimes nothing.
• Fields to Send: the columns written in both branches, mapped automatically or set one by one through Data to Send.

Changes the columns you map on a row that already exists. Columns left out of the mapper keep their current value, so a workflow can own one field without touching the rest of the line.

• Row ID Value: required here. NocoDB expects the ID of the record, not a name or a label, and the value usually comes from a previous search.
• Fields to Send: only the columns you want rewritten, with Inputs to Ignore as a safety net when the incoming item carries more properties than the table should receive.

Removes one record from the table. There is no confirmation step and no undo inside the workflow, so the node that feeds it matters more than the node itself.

• Row ID Value: the identifier of the record to remove, typically {{ $json.Id }} from a search run just before.
• Base Name or ID and Table Name or ID: worth double checking when several bases hold tables with the same name.

Fetches a single record by its identifier and returns its columns as one item, which is the cheapest way to refresh data you already have a pointer to.

• Row ID Value: the ID of the record to read.
• Download Field Names or IDs: the attachment columns to pull. The list only offers fields of type attachment.
• Download Attachments: off, you get the file metadata; on, n8n downloads the files named above as binary data, ready for an upload node or an email.

Returns the rows matching a filter, or the whole table when no filter is set. This is the operation a scheduled workflow builds on, since NocoDB has no trigger node.

• Return All: on, n8n walks every page of results; off, it stops at Limit.
• Filter By Formula: the NocoDB filter syntax, in the shape (name,like,example%)~or(name,eq,test).
• View Name or ID, Fields, Sort, Page and Shuffle: sit under Options to read through a saved view, trim the columns returned, order them or skip pages.

Returns how many rows match, without carrying the rows themselves through the workflow. Useful before a write, and cheap enough to run on a short schedule.

• Table Name or ID: the table to measure, in the base picked above.
• Filter By Formula: under Options, the same formula syntax as the search operation, which turns the count into a metric for one segment rather than the whole table.

Puts one or more files into an attachment cell of an existing row, so documents live next to the record instead of in a separate folder.

• Upload Mode: url to let NocoDB fetch the file from a public address, base64 to push the file content coming from an earlier node.
• Upload Field Name or ID: the attachment column receiving the file.
• Filename and Content Type: the name stored in NocoDB and the type, such as image/jpeg.
• Base64 Value or File Url: the file itself, depending on the mode.

Reads what a record points to through a link field, and gives back the related records as separate items. It answers the question a flat read cannot: what is attached to this line.

• Row ID Value: the record whose links you are reading.
• Link Field Name or ID: the relational column to follow, offered in the list for fields of type link.
• Return All and Limit: same pagination choice as elsewhere, with Fields, Sort and Filter By Formula available under Options.

Attaches one record to another through a relational field. The two rows already exist; this operation only draws the connection between them.

• Row ID Value: the record that holds the link field.
• Linked Row ID Value: the record being attached. Both are IDs, never names, which is the mistake that returns an error on the first run.
• Link Field Name or ID: the relational column that carries the relation.

Breaks the relation between two records without deleting either of them. The rows stay in their tables, only the connection goes.

• Link Field Name or ID: the relational column holding the relation to remove.
• Row ID Value and Linked Row ID Value: the two ends of the relation, in the same order as the link operation.

Lists the bases the token can reach, with their identifiers. Most workflows never need it, and the ones that do need it a lot: it is how a run discovers where to write instead of having the base hardcoded.

• Workspace Name or ID: optional, and the only setting here. Left empty, the node returns what the credential can see.

Retrieves a single base and its details, which is mostly a resolution step: you hold an identifier, you want the object behind it.

• Base Name or ID: the only required field, picked from the list or given by an expression when the identifier comes from an earlier node.