Installation

Install an official Lelu SDK, configure environment variables, and initialize the client in your app.

On this page

Choose Installation Method
Install the Package
CLI Commands
Set Environment Variables
Configure Lelu Client
That's it!

1Choose Installation Method

Lelu can be installed in two ways: using Docker (recommended for quick start) or by installing SDKs directly.

⚡ Recommended

One-Command Setup

Install SDK and automatically start all services with Docker. Includes engine, platform, UI, and database. Perfect for getting started quickly.

terminal

npm install lelu-agent-auth
npx lelu-agent-auth init

Opens browser to http://localhost:3002 when ready

Alternative Installation Methods

Manual Docker Setup

Pull and run Docker images manually. For users who prefer more control over the setup process.

terminal

# Clone and run all services locally
git clone https://github.com/lelu-auth/lelu.git
cd lelu
docker-compose up -d

SDK Only

Install the lightweight SDK with CLI tools for audit logs and policy management. Uses local SQLite storage by default (like Prisma).

terminal

# TypeScript/JavaScript
npm install lelu-agent-auth
npx lelu audit-log

# Python
pip install lelu-agent-auth-sdk
lelu audit-log

New to Lelu? We recommend starting with the one-command setup for the quickest start.

The SDK includes CLI tools for audit logs and policy management with local SQLite storage. For a visual UI, use Docker or visit lelu-ai.com.

2Install the Package

Let's start by adding Lelu to your project:

terminal

npm install lelu-agent-auth

If your frontend and backend are in separate repositories, install the Lelu SDK in each service that calls the Lelu Engine.

3CLI Commands

After installing, you can use the built-in CLI commands to view audit logs, manage policies, and launch the visual UI directly from your terminal:

Visual UI (Docker)

Launch the visual dashboard for managing policies and viewing audit logs. The UI runs as a separate Docker container (like Prisma Studio).

terminal

# Visit the hosted version
# https://lelu-ai.com/

# Or run locally with Docker
docker-compose up -d

TypeScript/Node.js

terminal

# View audit logs
npx @lelu-auth/lelu audit-log

# Manage policies
npx @lelu-auth/lelu policies list
npx @lelu-auth/lelu policies get auth
npx @lelu-auth/lelu policies set auth ./auth.rego

Python

terminal

# After installing: pip install lelu-agent-auth-sdk
lelu audit-log
lelu policies list
lelu policies get auth
lelu policies set auth ./auth.rego

Go

terminal

# Build and run CLI
cd sdk/go/cmd/lelu
go build -o lelu
./lelu audit-log
./lelu policies list

Note: CLI commands use local SQLite storage by default (~/.lelu/lelu.db).

To use a remote platform, set LELU_PLATFORM_URL environment variable.

4Generate API Key & Set Environment Variables

First, generate an API key to authenticate with the Lelu engine. Then create a .env file with your configuration.

1Generate API Key

API keys authenticate your requests and identify your tenant. Choose one of the following methods:

Method 1: PowerShell Script (Recommended for Self-Hosted)

powershell

# Generate and store API key automatically
./generate-api-key.ps1

# Your key will be:
# - Generated with secure random bytes
# - Stored in Redis
# - Added to your .env file automatically

2API Key

Your Lelu API key for authenticating requests to the engine. Generate a secure key using openssl rand -base64 32.

.env

LELU_API_KEY=your_secure_api_key_here

Security notice: Never commit your API keys to version control. Add .env to your .gitignore file.

5Configure Lelu Client

Initialize the Lelu client in your application to start authorizing agent actions:

lib/lelu.ts

import { LeluClient } from "@lelu-auth/lelu";

export const lelu = new LeluClient({
  baseUrl: process.env.LELU_ENGINE_URL!,
  apiKey: process.env.LELU_API_KEY!,
});

// Example: Authorize an agent action
const decision = await lelu.agentAuthorize({
  actor: "support_agent",
  action: "issue_refund",
  context: { confidence: 0.85 }
});

if (decision.requiresHumanReview) {
  console.log("Action queued for human approval");
} else if (decision.allowed) {
  console.log("Action approved autonomously");
}

For Python applications:

lelu_client.py

from lelu import LeluClient
import os

lelu = LeluClient(
    base_url=os.environ["LELU_ENGINE_URL"],
    api_key=os.environ["LELU_API_KEY"]
)

# Example: Authorize an agent action
decision = lelu.agent_authorize(
    actor="support_agent",
    action="issue_refund",
    confidence=0.85
)

if decision.requires_human_review:
    print("Action queued for human approval")
elif decision.allowed:
    print("Action approved autonomously")

✓That's it!

You're all set! Lelu is now configured and ready to use. Check out the following resources to learn more:

Previous: Introduction Next: Architecture