> ## Documentation Index
> Fetch the complete documentation index at: https://confidence-auth-testing.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart with the Confidence CLI

> Set up Confidence in your project and teach your AI assistant to manage flags, tracking, and documentation.

Get from an existing project to a working Confidence integration with one command. The wizard signs you in, installs the SDK for your framework, and configures Claude Code, Cursor, or Codex so you can keep working in your editor.

```bash theme={null}
npx @spotify-confidence/quickstart
```

Requires Node.js 24 or later. Run the command from the project you want to integrate.

## What You Can Set Up

Choose one or more goals in the wizard:

* **Feature flags.** Install the SDK and add a working flag example in your project.
* **Event tracking.** Instrument events so you can measure user behavior.
* **Session recordings (beta).** Capture user sessions.

The wizard also installs the Confidence plugin and connects the flags and docs tools. After setup, ask your assistant to:

* Create a flag, add variants, and set a targeting rule
* Add that flag to the checkout page in the repository
* Search the Confidence documentation and apply the recommended SDK pattern
* Migrate flags from PostHog, Eppo, Statsig, or Optimizely when the wizard detects an existing provider

For prompts that cover experiments, see [Use AI with Confidence](/docs/quickstarts/use-mcp).

## What the Wizard Does

1. Checks the required tools and detects your framework.
2. Signs you in to your Confidence workspace.
3. Installs the Confidence plugin for Claude Code, Cursor, or Codex.
4. Connects the Confidence flags and docs tools.
5. Asks which features to set up.
6. Shows the planned changes and waits for confirmation.
7. Uses your AI assistant to integrate the SDK and produce a working example.

The wizard detects React, Next.js, Node.js, Python, Swift, Kotlin, Java, and Go. If it cannot detect the framework, choose TypeScript, JavaScript, or another supported SDK.

## After Setup

Open the project in the IDE you configured and try these prompts:

> Create a flag called "checkout-redesign" and add it to the checkout page.
>
> Show the recommended Confidence SDK pattern for this project.

The assistant uses the plugin and the connected tools. Run the wizard again in the same directory if you skipped a step.

## Options

| Option | Description |
| - | - |
| `--dir <path>` | Run the wizard in a specific project directory |
| `--dry-run` | Preview the flow without making API calls |
| `--no-telemetry` | Turn off anonymous usage telemetry |

During project setup, the wizard asks before the AI assistant modifies files or installs packages. See the [CLI README](https://github.com/spotify/confidence-cli#security-note) for the permissions each assistant uses.

## Related Resources

<CardGroup cols={2}>
  <Card title="Use AI with Confidence" href="/docs/quickstarts/use-mcp">
    Connect MCP servers yourself and manage flags or experiments with prompts
  </Card>

  <Card title="SDK Integration" href="/docs/sdks/introduction">
    Integrate Confidence SDKs into your application
  </Card>

  <Card title="Configure a Flag" href="/docs/quickstarts/configure-flag">
    Create a feature flag in the Confidence UI
  </Card>

  <Card title="Migrate to Confidence" href="/docs/migrations/overview">
    Migrate from PostHog, Eppo, Statsig, or Optimizely
  </Card>
</CardGroup>