Field-type-aware serialization, decimal auto-normalization, human-readable validation errors, true batch API (up to 200 rows per request), upsert, lookup, automatic retry with exponential backoff, and parallel multi-table fetch.

Installation

Docker / Custom Nodes Directory

Copy the entire n8n-nodes-baserow-plus folder into your n8n custom nodes directory:

scp -r n8n-nodes-baserow-plus/ user@server:/path/to/n8n/custom/

Then restart n8n:

docker restart n8n n8n-worker

npm (when published)

cd ~/.n8n
npm install n8n-nodes-baserow-plus

Credentials

Create a new credential of type Baserow Plus API with:

| Field | Description | |-------|-------------| | Base URL | Your Baserow instance URL (e.g. https://api.baserow.io or https://baserow.example.com) | | API Token | Generate in Baserow under Settings → API Tokens |

The credential name (baserowPlusApi) is distinct from v2, so both can be installed side-by-side.

Operations

List

List rows with optional filters, sorting, full-text search, and server-side field filtering.

| Parameter | Description | |-----------|-------------| | Fetch All | Paginate through all rows automatically | | Page Size | Rows per page (max 200) | | Max Records | Stop after this many rows (0 = unlimited) | | Search | Full-text search across all text fields | | Include Fields | Comma-separated field names for server-side inclusion | | Exclude Fields | Comma-separated field names for server-side exclusion | | Sort | One or more sort rules (field + direction) | | Filters | One or more filter rules (field + operator + value) | | Select Fields | Client-side field selection (comma-separated) |

Get

Retrieve one or more rows by ID. Accepts a single ID or comma-separated IDs.

Create

Create a new row. Supports both auto-mapping from input JSON and manual field configuration.

Update

Update an existing row by ID. Leave Row ID blank to auto-detect from the input item's id field — works automatically after List, Get, or Lookup operations.

Delete

Delete a row by ID. Same auto-detect behaviour as Update.

Batch Create

Create multiple rows using Baserow's batch API endpoint. Sends up to 200 rows per request, automatically chunking larger sets.

Performance: A single batch API call replaces 200 individual HTTP requests. Creating 500 rows requires only 3 API calls instead of 500.

Batch Update

Update multiple rows in a single batch API call. Two modes:

1. Shared fields mode: Provide comma-separated Row IDs + field values applied to every row
2. Per-item mode: Leave Row IDs empty, enable Auto Map All, and include an id field in each input item

Batch Delete

Delete multiple rows in a single batch API call. Provide comma-separated Row IDs.

Upsert

Update if a matching row exists, otherwise create. Select a field to match on — the node searches for an existing row with that field value and updates it, or creates a new row if no match is found.

Lookup

Search for rows by field value. A standalone search operation that exposes what upsert does internally.

| Parameter | Description | |-----------|-------------| | Lookup Field | The field to search in | | Lookup Value | The value to search for | | Lookup Operator | equal, contains, or icontains | | Max Results | Maximum rows to return |

Multi-Table Fetch

Fetch data from multiple Baserow tables in parallel using a single shared filter field. All tables are queried concurrently via Promise.all.

Field-Type-Aware Serialization

The node fetches each table's field schema and uses field type information for precise serialization:

| Baserow Field Type | Serialization | |--------------------|---------------| | link_row | Integer array — [18] not '[18]'. Empty [] allowed. Single int auto-wrapped. Invalid values throw with a clear message. | | boolean | Coerces "true", "false", 1, 0 to actual booleans | | number | Coerces string to Number, then auto-rounds to the field's configured decimal places (reads number_decimal_places from Baserow schema). A field set to 0 decimals turns 12.874523 into 13, 2 decimals turns it into 12.87. No manual toFixed() needed. | | rating | Coerces to integer via Math.round() | | single_select | Ensures string | | multiple_select | Ensures string array (accepts comma-separated strings) | | date | Ensures ISO string | | text, long_text, url, email, phone_number | Ensures string (stringifies objects) | | file | Pass-through | | Unknown | Primitives pass through, objects stringify |

When field type info is unavailable (e.g. cache miss), falls back to a heuristic that preserves backward compatibility with v2.

Error Handling & Retry

• Human-readable validation errors: Field IDs are translated to column names using the schema cache. Instead of {"field_267": ["A valid number is required."]} you get "Revenue": A valid number is required. Multiple field errors are piped together with |.
• Batch validation errors deduplicated: When the same field has the same error across multiple rows, it collapses to All rows / "tech_score": Ensure that there are no more than 0 decimal places. instead of repeating per row. Unique errors per row are shown individually as Row 0,2 / "field": message.
• 429 retry: Automatic retry on rate limiting with exponential backoff (1s, 2s, 4s). Respects Retry-After header.
• Configurable timeout: Default 30s, adjustable per-node via the "Request Timeout" parameter
• Continue on fail: All operations respect n8n's continueOnFail() pattern

Bug Fixes from v2

1. [null] → [0] in sanitizeWriteValue — Number(null)===0 passed integer check. Fixed with null/undefined guard.
2. Empty array [] → "[]" — Now passes through as [] for link_row fields.
3. Multi-select vs linked record overlap — Both checked value[0]?.value. Now checks for .color property (select options have {id, value, color}, linked records have {id, value} only).
4. Upsert matchValue broken with autoMapAll — Convoluted originalFields lookup failed. Now reads directly from items[i].json[upsertField].
5. minFilterValue rejects 0 — Used !== '' which rejected 0. Now uses != null.
6. batchCreate not batched — Was identical to create (1 API call per row). Now uses Baserow's real batch endpoint.

Configuration

| Setting | Default | Description | |---------|---------|-------------| | Use Field Names Mapping | true | Convert between field_xxx IDs and human-readable names | | Return Raw | false | Skip all name conversion and value cleaning | | Refresh Field Schema | false | Bypass the 5-minute schema cache | | Request Timeout | 30000 ms | Per-request timeout | | Auto Map All Input Fields | true | Map all incoming JSON keys automatically |

License

MIT

Author

Steven Lefebvre — expertlocal.ca