Superset for VS Code

Work with Apache Superset from inside VS Code: author Jinja-SQL, run queries, browse databases and saved queries, visualize object lineage, and spin up throwaway dataset experiments — without leaving your editor.

Status: early (0.0.1). The SQL/data workflow is the focus; chart and dashboard rendering stays in Superset's own UI (the extension deep-links to it).

Features

Connections — configure multiple Superset servers; passwords are stored in VS Code SecretStorage, never in settings.
Jinja-SQL editing — dedicated language mode with syntax highlighting, completions, hover docs, and diagnostics for Superset's Jinja templating.
Query execution — run the file or a selection (Cmd+Enter), estimate cost, and format SQL server-side.
Results — sortable, paginated results panel with Copy and Export CSV, plus an aligned text preview in the Superset output channel.
Object browser — tree views for Connections (database → schema → table), Saved Queries (open / create / update / delete), and Dashboards.
Dependency graph — an interactive lineage view: Database → Schema → Dataset → Chart → Dashboard. Click any node to trace its full upstream + downstream; filter by type and object; click through to the browser.
Deep links — open charts and dashboards in your browser, or inside VS Code's Simple Browser (superset.viewIn).
Experiment sandbox — clone a virtual (SQL) dataset's whole downstream lineage (dataset + charts + dashboards) into an isolated experiment, then tear it all down in one click. Originals are never modified.

Requirements

VS Code ^1.95
A reachable Apache Superset instance (local or remote) and an account on it

Don't have one yet? The quickest local setup is Docker:

docker run -d -p 8088:8088 --name superset apache/superset
docker exec -it superset superset fab create-admin \
  --username admin --firstname admin --lastname admin --email admin@local --password admin
docker exec -it superset superset db upgrade
docker exec -it superset superset init
docker exec -it superset superset load_examples   # optional sample data

Superset is then at http://localhost:8088 (admin / admin).

Install

Not yet published to the Marketplace — build from source:

git clone https://github.com/painterV/vscode-superset.git
cd vscode-superset
npm install
npm run compile

Then press F5 in VS Code to launch an Extension Development Host with the extension loaded. To produce an installable .vsix:

npm install -g @vscode/vsce
vsce package
code --install-extension vscode-superset-0.0.1.vsix

Getting started

Add a connection — open Settings (JSON) and add:

"superset.connections": [
  { "name": "Local", "url": "http://localhost:8088", "username": "admin" }
]

Connect — Cmd+Shift+P → Superset: Connect, pick the connection, enter your password (optionally remembered in SecretStorage).
Run a query — open a .jinjasql file (or switch a file's language to Jinja SQL), click a database in the Connections view to activate it, then Cmd+Enter.
Explore — open Superset: Dependency Graph from the Dashboards view to see lineage; use the Experiments view to clone a dataset for safe experimentation.

Commands

Command	Description
`Superset: Connect` / `Disconnect` / `Switch Connection`	Manage the active connection
`Superset: Run Query`	Execute the file or selection (`Cmd+Enter`)
`Superset: Estimate Query Cost`	Estimate cost (`Cmd+Shift+E`)
`Superset: Format SQL`	Server-side SQL formatting (`Cmd+Shift+F`)
`Superset: Save as Saved Query`	Persist the current query to Superset
`Superset: Dependency Graph`	Open the lineage graph
`Superset: Clone Dataset as Experiment`	Clone a virtual dataset's lineage into a sandbox
`Superset: Refresh`	Refresh the tree views

Charts/dashboards offer Open in Browser actions; experiments offer Delete Experiment.

Settings

Setting	Default	Description
`superset.connections`	`[]`	List of `{ name, url, username }` connection configs
`superset.defaultDatabase`	`""`	Default database name for query execution
`superset.resultPageSize`	`50`	Rows per page in the results panel
`superset.maxResultTabs`	`5`	Max result tabs kept in memory
`superset.schemaCacheTtlSeconds`	`300`	Cache TTL for schema/table lists
`superset.queryTimeoutSeconds`	`30`	Show a progress notification after this many seconds
`superset.viewIn`	`"browser"`	Open charts/dashboards in your `browser` or VS Code's `editor` (Simple Browser)

editor mode requires TALISMAN_ENABLED = False in your superset_config.py (Superset blocks framing by default) and a login inside the embedded view.

Development

npm run watch      # rebuild on save (esbuild + tsc)
npm run lint       # eslint
npm test           # unit tests (downloads a throwaway VS Code via @vscode/test-electron)

Source layout:

src/
  api/            SupersetClient + typed endpoint wrappers (auth, sqllab, charts, …)
  language/       Jinja-SQL completion / hover / diagnostics providers
  views/          tree views + webview panels (results, lineage, experiments)
  experiments/    dataset experiment clone + teardown
media/            webview assets (CSS/JS for results & lineage)
test/unit/        mocha unit tests
docs/superpowers/ design specs and implementation plans

How it works (briefly)

The extension talks to Superset's REST API (/api/v1/...) over fetch. Auth uses the JWT login endpoint plus the CSRF token and its session cookie. Chart and dashboard visuals are intentionally not re-rendered in VS Code — a Superset chart is a live, server-rendered visualization, so the extension deep-links to Superset's UI for those and keeps its focus on the SQL/data side, where an editor is genuinely better.

Contributing

Issues and PRs are welcome. Please run npm run lint and npm test before opening a PR, and keep changes focused.

Superset Studio

Wenbaoli