Modelith for VS Code

Ontology-anchored, git-native data modeling for dbt teams, inside VS Code, standalone or in a devcontainer. Model your warehouse as an entity-relationship diagram beside your YAML, generate contract-enforced dbt, and catch drift, without leaving the editor.

Prerequisite

This extension drives the mdl command-line tool. Install it first:

uv tool install modelith-dbt
# or: pipx install modelith-dbt

The extension does not bundle its own copy of the canvas. It launches mdl serve and embeds the live canvas, so whatever the CLI understands, the editor shows.

Install the extension from the Marketplace UI, or by id:

code --install-extension BosonResearch.modelith-vscode
# cursor --install-extension BosonResearch.modelith-vscode   # Cursor / Windsurf / VSCodium (Open VSX)

What you get

• Canvas beside your YAML. Right-click a model file and choose Modelith: Open Model Preview to the Side. The canvas opens in a split beside your editor and follows the active file: open an entity's YAML and the diagram centers on it, switch to a generated .sql file and the preview tracks it, save the YAML and it re-renders. You read and edit the model as text on one side and watch the diagram update on the other.
• The full editable canvas. Modelith: Open Canvas opens the complete editor as a tab: drag between entities to draw relationships, edit attributes and named keys inline, browse the ontology and the four-layer stack, and commit from a git panel. modelith.canvas.display picks a webview tab or an external browser tab via a forwarded port, so devcontainers and remote work out of the box.
• Diagnostics on save. mdl validate runs when you save a model YAML and surfaces MDL-* findings in the Problems panel, mapped to the file that declares the issue. The status bar shows a valid or error-count chip.
• Language server. Drift and contract diagnostics land on the generated dbt files, with hover cards (glossary term, ontology IRI, owner) and code actions (adopt a column, lift a model, unmanage, declare a relationship).
• Commands. Generate the dbt project, check drift, lint and fix naming, add an entity, vendor an ontology, emit the semantic layer, all from the command palette.
• YAML completion. JSON Schemas exported from the model are registered with the Red Hat YAML extension, so authoring the YAML by hand is schema-checked and autocompleted.
• mdl detection. Resolves in order: an explicit modelith.mdlPath setting, a project .venv, mdl on PATH, the active conda or virtualenv bin, then common per-user install locations. A standard uv tool install modelith-dbt needs no configuration. If nothing matches, run which mdl in the integrated terminal and set that as modelith.mdlPath.

AI assistance (Copilot Chat)

Modelith registers two AI integrations, one per Copilot Chat mode. Both use the same mdl engine already detected for everything else — no mcp.json to write, no separate install.

• Agent mode — Modelith MCP tools. In Copilot Chat's Agent mode the model is available as tools: list_entities, get_entity, search_ontology, get_model_context, validate, and the write tools create_entity / update_entity. The agent can ground itself in what already exists, search the ontology for an alignment, and write validated entities straight into your checkout (direct-write — you review the git diff, the same trust boundary as the CLI). The tools only run in Agent mode: tool invocation needs the plan/act/observe loop that Ask mode does not have. The server is the mdl mcp subcommand, registered automatically on activation and scoped to your workspace model. Requires VS Code 1.99+ (older hosts keep every other feature; only the tools are absent).
• Ask mode — @modelith. In Ask mode, where MCP tools cannot run, type @modelith and ask about the model: name an entity for its attributes / ontology alignment / relationships, @modelith /list for the entity list, @modelith /explain <entity> for one in full, or a general question for a model summary. It reads the model through the same query layer the MCP tools use, so the two modes answer consistently.

Devcontainers

The extension declares "extensionKind": ["workspace"], so in a devcontainer it runs inside the container, next to mdl, dbt, and your warehouse credentials. The canvas server binds in the container and VS Code forwards the port automatically. Two ready-made setups ship in the Modelith repository: a minimal template under profiles/devcontainer/, and a full onboarding devcontainer at .devcontainer/ that installs the CLI, scaffolds a demo model, and opens the ER canvas on first attach.

Learn more

Modelith is a full data-modeling toolchain (a CLI, a web canvas, a language server, reverse engineering, drift detection, and governance sync). See the project repository for the complete picture.