databricks-tui

Terminal dashboard for Databricks — monitor compute, jobs, pipelines, SQL warehouses, dashboards, and Unity Catalog in one view.

Community project — not affiliated with, endorsed by, or supported by Databricks, Inc. It drives the official Databricks CLI with your own credentials; nothing is sent anywhere except your workspace.

• Animated splash screen and a Databricks-branded look: product-named panes (Compute, Lakeflow, SQL Warehouses, AI/BI Dashboards), status chips, and refresh flashes when new data lands
• Five color-coded panes that populate independently as each data source responds
• Jobs show their latest run result and a ✓✗✓ history strip, not just the config
• Health summary in the header: running / pending / failed / idle counts at a glance
• Active work floats to the top of every pane: running clusters, jobs and warehouses first, then pending, failures, finished, idle
• Grep any pane with /: filter by name, detail text or status (/running shows only running things)
• Drill into any item: formatted key facts and recent activity, raw JSON one key away
• Debug Lakeflow without the browser: drill from a job into any run (per-task states, durations, the actual error output of failed tasks) or from a pipeline into any update (state, cause, event-log entries with failures highlighted); live runs refresh themselves every few seconds
• SQL console (:): type a statement, run it on a warehouse, page through results — the whole lakehouse is queryable from the dashboard; with a table/view selected in the catalog pane, the prompt opens pre-filled with SELECT * FROM catalog.schema.table LIMIT 100, ready to edit; ↑/↓ cycle through past statements (persisted in ~/.config/databricks-tui/history), Ctrl+R searches them incrementally, Ctrl+X composes multi-line SQL in your $EDITOR, and Ctrl+S exports results to CSV — previews export with e
• Problems view (!): everything currently failing across all panes in one list, with Enter jumping straight to the culprit — and when a new failure appears between refreshes, the footer flashes it (with a terminal bell) the moment it happens
• Command palette (Ctrl+P): fuzzy-search every loaded resource across panes and jump straight to it
• Cancel from where you stand: s in a run/update view cancels it, and Esc on a running SQL statement cancels it server-side instead of leaving it burning the warehouse
• Act on resources: start/stop clusters, warehouses and pipelines, trigger job runs
• Inspect access: effective Unity Catalog grants (with inheritance) and workspace object ACLs for any selected item
• Warehouse details include recent query history: who ran what, how long
• DBU usage view scoped to the current workspace: 14 days of system.billing.usage as color-stacked daily bars, bucketed by SKU family (Jobs / SQL / All-Purpose / DLT), with list-price dollar estimates when system.billing.list_prices is readable, and a top-spenders table naming the jobs/clusters/warehouses behind the DBUs
• Table lineage as a tree: up to 3 hops upstream and downstream from system.access.table_lineage, rendered with branch guides so you can follow a table's ancestry at a glance
• Jump to any resource in the workspace web UI with one key
• Browse Lakeview dashboards: pages, widgets and datasets at a glance
• Unity Catalog browser: drill from catalogs into schemas, tables, views and volumes — and into the volumes themselves, browsing files and directories with sizes and ages, with Enter showing the head of any text file; table details include the full column schema plus size, file count, format and last-modified from DESCRIBE DETAIL, and p previews sample rows
• Arrange the dashboard your way: H hides and reorders panes, the grid adapts to what's visible, and the arrangement persists across sessions
• ? shows every keybinding, grouped by context, lazygit-style
• Secret scopes pane: browse scopes and keys (with last-updated), create scopes, add secrets (values masked while typing, never displayed), delete with confirmation, and inspect scope ACLs with g
• Switch between workspaces (CLI profiles) without restarting
• Zoom into any pane, non-blocking refresh — the UI never freezes
• Eight color themes: terminal-default dark, light, Catppuccin Mocha & Latte, Gruvbox, Dracula, Nord and Tokyo Night — t cycles, --theme picks at launch, and the app remembers your theme and warehouse choice per profile across sessions (~/.config/databricks-tui/config.json)
• Built-in self-upgrade from GitHub releases

Install

With Homebrew (macOS and Linux):

brew install pjhamera/tap/databricks-tui

Or download the latest release for your platform from the releases page:

# macOS (Apple Silicon)
curl -sL https://github.com/pjhamera/databricks-tui/releases/latest/download/databricks-tui-macos-arm64.tar.gz | tar xz
mv databricks-tui /usr/local/bin/

Artifacts: databricks-tui-macos-arm64, databricks-tui-macos-x86_64, databricks-tui-linux-x86_64 — each with a .sha256 checksum.

Or with cargo:

cargo install databricks-tui

Upgrade

databricks-tui upgrade

Detects your platform, checks the latest GitHub release, and replaces the binary in place if a newer version exists.

Uninstall

databricks-tui uninstall          # asks for confirmation
databricks-tui uninstall --yes    # no prompt

Removes the binary from wherever it is installed. The only other files the app keeps are under ~/.config/databricks-tui/ (SQL history, preferences).

Usage

databricks-tui                      # default profile, 30s refresh
databricks-tui --profile prod       # named CLI profile
databricks-tui --refresh 10         # refresh every 10 seconds
databricks-tui --theme light        # or catppuccin-mocha, catppuccin-latte,
                                    # gruvbox, dracula, nord, tokyo-night

The Clusters pane shows interactive (UI/API-created) clusters only — job-created clusters are excluded, both for signal and because listing them can be slow on busy workspaces.

Keys

Navigation works while zoomed — Tab/h/l jumps straight to the next panel full-screen. In the details view, j/k scroll, J toggles the raw JSON, o opens the browser, and Esc goes back.

Requirements

• Databricks CLI v0.200+ installed and authenticated

See docs/permissions.md for what each feature needs (system tables, warehouse access, …) and docs/troubleshooting.md for common issues.

Release binaries

Push a v* tag to trigger a GitHub Actions build that publishes .tar.gz binaries (with sha256 checksums and auto-generated release notes) for Linux x86_64, macOS x86_64, and macOS ARM.