Cluster Setup

Alpha Access Required

Lucid is in private alpha. Request access to get started.

This guide covers how to connect your AI workloads to the Lucid platform for TEE-based security.

Using Lucid-Managed Infrastructure (Recommended)

The simplest way to use Lucid is through the hosted platform, where TEE infrastructure is managed for you.

Step 1: Authenticate

lucid loginAuthenticated as user@example.com

Step 2: Deploy from YAML

Create a my-env.yaml file:

apiVersion: lucid.io/v1alpha1
kind: LucidEnvironment
metadata:
  name: my-platform
spec:
  infrastructure:
    provider: gcp
    region: us-central1
  agents:
    - name: my-agent
      model:
        id: meta-llama/Llama-3.3-70B
      gpu:
        type: H100
        memory: 80GB

Deploy it:

lucid apply -f my-env.yamlAgent created: agent-abc123

Step 3: Verify Status

lucid status my-agentAgent: my-agent
ID: agent-abc123
Status: running
Model: meta-llama/Llama-3.3-70B
GPU: H100

That's it! Your agent is running on TEE-capable hardware managed by Lucid.

---

Self-Hosted Deployment (Advanced)

For users who want full control, deploy to your own Kubernetes cluster using lucid apply.

Prerequisites

• Kubernetes cluster (GKE, EKS, AKS, or local Kind/minikube)
• kubectl configured and connected to your cluster
• TEE-capable nodes (for production) or mock mode (for development)

Provider	TEE Requirement
GCP	GKE with Confidential GKE Nodes (AMD SEV-SNP)
Azure	AKS with DCsv3 or ECsv3 nodes (Intel SGX)
AWS	EKS with Nitro Enclaves enabled
Local	Kind or minikube with mock TEE mode (see Local Development)

Step 1: Create Environment File

Create a my-env.yaml file:

apiVersion: lucid.io/v1alpha1
kind: LucidEnvironment
metadata:
  name: my-platform
spec:
  infrastructure:
    provider: gcp                    # gcp | aws | azure | local
    region: us-central1
    projectId: my-gcp-project
    confidentialComputing:
      enabled: true
    cluster:
      name: lucid-cluster
      nodePools:
        - name: gpu-pool
          machineType: a3-highgpu-8g
          gpuType: nvidia-h100-80gb
          gpuCount: 8

  agents:
    - name: my-assistant
      model:
        id: meta-llama/Llama-3.3-70B
      gpu:
        type: H100
        memory: 80GB

  services:
    observability:
      enabled: true

Step 2: Preview Deployment

lucid diff -f my-env.yamlEnvironment: my-platform

Infrastructure:
Provider: gcp
Region: us-central1
Cluster: lucid-cluster
CC Mode: True

Agents (1):
- my-assistant [enabled]
Model: meta-llama/Llama-3.3-70B
GPU: H100 (80GB)

Run 'lucid apply -f <file>' to deploy.

Step 3: Deploy

lucid apply -f my-env.yamlThis will deploy:
- Infrastructure: gcp / us-central1
- Cluster: lucid-cluster
- 1 agent(s)
- Will provision GCP resources (this may incur costs)
Proceed? [y/N]: y

==================================================
Step 1: Infrastructure Provisioning
==================================================
[*] Provisioning GCP infrastructure...
[+] Infrastructure provisioned: lucid-cluster

==================================================
Step 2: Cluster Setup
==================================================
[*] Checking operator at http://localhost:8443...
[!] Lucid operator not found on cluster
Install the Lucid operator now? [y/N]: y
[*] Running: helm install lucid-operator oci://us-central1-docker.pkg.dev/lucid-prod/lucid-charts/lucid-operator -n lucid-system --create-namespace --wait --timeout 5m
[+] Operator installed successfully
[*] Waiting for operator to become ready...
[+] Operator is healthy

==================================================
Agent Deployment
==================================================
Creating agent: my-assistant...
Created: agent-abc123

==================================================
Deployment Summary
==================================================
Infrastructure: provisioned
Cluster: installed
Agents: 1 deployed

[+] Environment deployed successfully!

Flags

Flag	Description
--skip-infra	Use existing cluster, skip infrastructure provisioning
-y, --yes	Skip confirmation prompts
--managed	Use Lucid managed deployment

---

Local Development (Mock Mode)

For local development and testing, you can set up a local Kind cluster with mock TEE mode.

Setup with Kind

First, create a Kind cluster:

# Create a Kind clusterkind create cluster --name lucid-localCreating cluster "lucid-local" ...
✓ Cluster ready
# Verify kubectl contextkubectl cluster-info --context kind-lucid-localKubernetes control plane is running at https://127.0.0.1:PORT

Then create a local environment file (local-env.yaml):

apiVersion: lucid.io/v1alpha1
kind: LucidEnvironment
metadata:
  name: local-dev
spec:
  infrastructure:
    provider: local
  agents:
    - name: test-agent
      model:
        id: meta-llama/Llama-3.3-70B
      gpu:
        type: mock  # Uses mock GPU for local development

Deploy it:

lucid apply -f local-env.yaml -y[+] Environment deployed successfully!

Observer UI: http://localhost:3000
Verifier API: http://localhost:8000

The operator URL is auto-detected from LUCID_OPERATOR_URL env var or defaults to localhost:8443.

Mock Mode provides:

• Identical API contracts as production
• Software-based attestation signatures
• Full auditor chain functionality
• Safe development environment

Code Portability

Code developed in Mock Mode works unchanged when deployed to production TEE environments.

---

Next Steps

Once your cluster is ready:

1. Build Your First Auditor - Create a custom safety auditor
2. Deployment Modes Guide - Learn about CLI vs GUI deployment
3. View AI Passports - Verify attestation results