VS Code extension 2.21.0 made the direct engine the only bundle deployment engine

The Databricks extension for VS Code shipped four releases in October 2026. Version 2.20.0, on October 1, raised the minimum supported VS Code version to 1.104 and made the Bundle Variables view refresh immediately when overrides are saved or reset. Its upgraded CLI integration migrated bundle deployment state to the direct engine and returns an error if a configured workspace ID does not match the connected workspace [6].

The largest change came in v2.21.0 on October 12. Databricks Asset Bundles now deploy only with the direct engine, and the Terraform deployment engine and its bundled binary were removed. The same release restored default TLS certificate verification for SDK calls and sorted folders alphabetically in the Workspace File System view [4].

Version 2.22.0, on October 22, stopped bundle sub-project discovery from blocking the extension host and had VS Code handle proxies and certificates for Databricks SDK calls. It fixed environment selection and constraint conflicts for uv workspace members, and it exposed the bundle Configuration view and Bundle Resource Explorer in Remote SSH mode [3]. Version 2.23.0, on October 29, added the databricks.notebooks.configureJupyterCellMarkers setting, which stops the extension from writing Jupyter cell-marker settings to workspace settings. It also made claude and codex available in SSH tunnel connections with Unity Gateway preconfigured, and it bundled Databricks CLI v1.23.0 [1].

In the community, a thread on October 7 asked about maintaining Databricks Genie agents and whether a VS Code extension was available for that work [5]. Separately, a dbt v2 article on October 27 described a Rust-based Jinja engine and much faster local linting [2].