Quickstart
Prerequisites
- Databricks CLI
v1.0.0+with an authenticated profile - Node.js 22.16+ (AppKit apps are Node/TypeScript)
- pnpm, which the default AppKit template uses to install dependencies and run scripts
- Databricks workspace with Apps enabled
- For the manual path, a Lakebase Postgres project you have
CAN MANAGEon, with its branch and database
Template path
Templates are agent-ready prompts organized by use case. Pick one that fits, copy it into your AI coding assistant, and the assistant handles scaffolding, plugin selection, and deployment.
Common starting points:
| Template | Best for |
|---|---|
| Set Up Your Local Dev Environment | Install the CLI, authenticate, verify the workspace |
| Spin Up a Databricks App | Scaffold a fresh AppKit app, run it locally, and deploy |
| Onboard Your Coding Agent | Install agent skills, wire up the DevHub Docs MCP server |
| AI Chat App | Conversational AI, chatbots, assistants |
| App with Lakebase | CRUD apps with persistent storage |
The templates catalog has the full list, including Lakebase Postgres, Genie Agents, Unity Gateway, and Agent Bricks.
Give your AI assistant Databricks platform context by installing the agent skills before copying in the template:
databricks aitools installManual path
Without a template, databricks apps init generates a working AppKit project. With --features lakebase, it generates a sample to-do app on an app.todos table. Here is a shortened version of the server code it produces (you don't write this yourself):
import { createApp, lakebase, server } from "@databricks/appkit";
import { setupSampleLakebaseRoutes } from "./routes/lakebase/todo-routes";
createApp({
plugins: [lakebase(), server()],
async onPluginsReady(appkit) {
await setupSampleLakebaseRoutes(appkit);
},
}).catch(console.error);export async function setupSampleLakebaseRoutes(appkit) {
// On first start, creates the app schema and the app.todos table
await appkit.lakebase.query("CREATE SCHEMA IF NOT EXISTS app");
await appkit.lakebase.query(
"CREATE TABLE IF NOT EXISTS app.todos (id SERIAL PRIMARY KEY, title TEXT NOT NULL, ...)",
);
appkit.server.extend((app) => {
app.get("/api/lakebase/todos", async (_req, res) => {
const { rows } = await appkit.lakebase.query(
"SELECT id, title, completed, created_at FROM app.todos ORDER BY created_at DESC",
);
res.json(rows);
});
});
}Scaffold, deploy, and run locally. Deploy before you run locally so the app's service principal creates and owns the database schema; Local setup explains why.
# Generate the project and install dependencies
databricks apps init --name my-app --features lakebase \
--set lakebase.postgres.project=projects/<project-id> \
--set lakebase.postgres.branch=projects/<project-id>/branches/<branch-id> \
--set lakebase.postgres.database=projects/<project-id>/branches/<branch-id>/databases/<database-id>
cd my-app && databricks apps deploy # deploy to your workspace
pnpm run dev # run locallyAfter deploy, the CLI prints your app's workspace URL.
To scaffold with specific plugins, pass --features with a comma-separated list. Run databricks apps manifest to see all available plugins and their required resource fields.
Where to next
For the full local development workflow, deploy flags, and plugin setup, see Apps development.