> For the complete documentation index, see [llms.txt](https://docs.n8n.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.vectorstoredatabricks/migrate-from-the-community-package.md). # Migrate from the community package Move workflows from the n8n-nodes-databricks community package to the built-in Databricks nodes: node mapping, field differences, and behavior changes. The built-in Databricks nodes replace the [n8n-nodes-databricks](https://www.npmjs.com/package/n8n-nodes-databricks) community package. This page maps each community node to its built-in counterpart, lists the fields that moved, and describes the behavior changes to expect. Community nodes and built-in nodes have different node types, so you can't convert a node in place. For each community node in a workflow: add the built-in node, copy the settings over, reconnect it, and delete the community node. The built-in AI nodes need a [Databricks OAuth2 credential](/integrations/builtin/credentials/databricks.md#supported-authentication-methods) with a service principal or user login. The community package's host and personal access token credential has no built-in equivalent for the AI nodes; only the built-in [Databricks](/integrations/builtin/app-nodes/n8n-nodes-base.databricks.md) and [Databricks Trigger](/integrations/builtin/trigger-nodes/n8n-nodes-base.databrickstrigger.md) nodes accept a personal access token. ## Node mapping | Community node | Built-in node | What changed | | ----------------------- | ---------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Databricks | [Databricks](/integrations/builtin/app-nodes/n8n-nodes-base.databricks.md) | Same resources. The built-in node adds Job operations and a [Databricks Trigger](/integrations/builtin/trigger-nodes/n8n-nodes-base.databrickstrigger.md) node. | | Databricks Chat Model | [Databricks Chat Model](/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.lmchatdatabricks.md) | **Model** selects a Unity AI Gateway model service, `catalog.schema.service`, instead of a serving endpoint. The **Conversation ID** and **Return Trace** options don't exist. | | Embeddings Databricks | [Embeddings Databricks](/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.embeddingsdatabricks.md) | **Serving Endpoint** becomes **Model**, a list of the embeddings-capable model services. | | Databricks Vector Store | [Databricks Vector Store](/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.vectorstoredatabricks.md) | **Index Name** becomes **Index**, a list or the full `catalog.schema.index` name. **Text Column** becomes **Content Column**, picked from the index's columns. **Metadata Columns** is a multi-select instead of a comma-separated list. **Score Threshold** is gone; use **Limit**. New: **Search Mode** and **Metadata Filter**. | | Databricks AI Agent | [Databricks AI Agent](/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.databricksaiagent.md) | **Enable MLflow Tracking** becomes **Export Traces to MLflow**. The default experiment path `/Shared/n8n-workflows-` stays the same, and the new **Experiment** field lets you pick an existing experiment. The credential on the agent is only needed for the export. | ## Field differences on the Vector Store node | Community field | Built-in field | Notes | | ------------------------------------------- | --------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Index Name** (text) | **Index** | Pick **From List**, or enter the full name in **Name**. In **Insert Documents** mode the list shows Direct Access indexes only. | | **Text Column** (default `text`) | **Content Column** | Picked from the index's columns. Leave it empty on a managed index to use the embedding source column. | | **Metadata Columns** (comma-separated text) | **Metadata Columns** option | A multi-select. Empty returns every column except the primary key, the content column, and the vector column. | | **Score Threshold** (default `0.8`) | None | Use **Limit**. Refer to [No score threshold](/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.vectorstoredatabricks.md#no-score-threshold) for why. | | None | **Search Mode** option | **Vector** or **Hybrid**. | | None | **Metadata Filter** option | A JSON filter sent as `filters_json`. | ## Behavior changes to expect * **Searches on managed indexes work.** The community node always sent `query_vector`, so every search on a Delta Sync index with Databricks-computed embeddings failed. The built-in node sends `query_text` or `query_vector` as the index requires. Refer to [Index types and embeddings](/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.vectorstoredatabricks.md#index-types-and-embeddings). * **No results dropped by a threshold.** The community node's **Score Threshold** defaulted to `0.8`, which discarded every result on indexes whose similarity scores sit far below that. The built-in node returns the top **Limit** results. * **Hybrid search and metadata filters** are available in all search modes, including when an agent uses the index as a tool. * **Inserts are checked before the request.** A Delta Sync index or a wrong **Content Column** fails with a message that names the problem instead of a raw API error. * **The agent doesn't drift.** The built-in Databricks AI Agent runs the AI Agent node's own code instead of a fork, so it has the same options and gets the same fixes. * **Traces keep landing in the same experiment** when you replace the agent node inside an existing workflow, because the default experiment path contains the workflow ID. ## Related resources * [Databricks Vector Store](/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.vectorstoredatabricks.md) * [Embeddings Databricks](/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.embeddingsdatabricks.md) * [Databricks AI Agent](/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.databricksaiagent.md) * [Databricks Chat Model](/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.lmchatdatabricks.md) * [Databricks credentials](/integrations/builtin/credentials/databricks.md)