Part of digitaleraluhut — the application layer. Each app deploys onto the homelab cluster and connects to local-ai for inference, transcription, and image generation.

Personal homelab app deployments — each app is a Pulumi project that consumes @mrsimpson/homelab-core-components and deploys onto the homelab k3s cluster.

homelab-apps/
├── package.json               ← npm workspaces root; shared devDeps (typescript, @types/node)
├── tsconfig.base.json         ← shared TypeScript compiler options
├── apps/
│   └── lobehub/               ← self-hosted LobeHub (AI chat)
│       ├── Pulumi.yaml
│       ├── Pulumi.dev.yaml
│       ├── src/
│       │   ├── index.ts
│       │   └── models.ts
│       ├── package.json
│       └── tsconfig.json
└── .github/
    └── workflows/
        └── deploy-lobehub.yml ← calls digitaleraluhut/homelab/.github/workflows/deploy-to-cluster.yml

Each app in apps/<name>/ is a standalone Pulumi project. It reads shared infrastructure outputs (domain, tunnel CNAME, Cloudflare zone) via a StackReference to your homelab base stack and manages its own Kubernetes resources in an isolated Pulumi state.

GitHub Actions deploys to the cluster by calling the reusable deploy-to-cluster.yml workflow from the digitaleraluhut/homelab repo.

These apps are designed to be forkable. They depend on a separate homelab base stack that provides shared infrastructure (Cloudflare tunnel, domain, zone ID). You need to run that first.

1. A running k3s (or compatible) cluster
2. The homelab base stack deployed and its Pulumi stack outputs available (exports domain, cloudflareZoneId, tunnelId)
3. A Pulumi Cloud account (free tier works)
4. pulumi CLI authenticated: pulumi login

cd apps/<name>

# Create your stack
pulumi stack init dev

# Point it at YOUR homelab base stack (replace with your Pulumi org)
pulumi config set <app>:homelabStack <your-pulumi-org>/homelab/dev

# Copy the example config and fill in your values
cp Pulumi.dev.yaml.example Pulumi.dev.yaml
# Edit Pulumi.dev.yaml — replace all <placeholder> values
# Then import it:
pulumi config --path < Pulumi.dev.yaml

# Set required secrets (see Pulumi.dev.yaml.example comments for what's needed)
pulumi config set <app>:someSecret <value> --secret

# Deploy
pulumi up

Note: homelabStack has no default — Pulumi will error immediately if it is not set, preventing silent cross-org stack references.

Fork both repos (homelab and this one), then add these secrets to your fork of homelab-apps:

The deploy workflows call a reusable deploy-to-cluster.yml from the homelab repo — update the uses: line in each workflow file to point to your fork.

# 1. Authenticate with Pulumi Cloud (restores all encrypted stack config)
pulumi login

# 2. For each app, select the stack — Pulumi pulls config from the cloud automatically
cd apps/<name>
pulumi stack select dev

# 3. Fill in plaintext values (node IP, hostname, etc.) from the example file
cp Pulumi.dev.yaml.example Pulumi.dev.yaml
# edit Pulumi.dev.yaml — replace <placeholders> with your values
# e.g.: pulumi config set aftertouch:nodeIp 192.168.x.x --stack dev

Bootstrap all secrets in one step from the digitaleraluhut/homelab repo:

./scripts/setup-homelab-apps.sh

This repo ships agentskills.io-compatible skills under skills/. Load them in your AI agent to deploy and configure apps without reading documentation manually.

Load the deploy-homelab-app skill in your AI agent and say "add a new app to homelab-apps".

Or manually:

1. Create apps/<name>/ following the structure of apps/lobehub/
2. Add a deploy workflow .github/workflows/deploy-<name>.yml
3. Run pulumi up locally once to create the Kubernetes namespace
4. Generate a namespace-scoped KUBECONFIG:

    # from the digitaleraluhut/homelab repo
    bash scripts/create-kubeconfig.sh <name>
    base64 -w0 /tmp/<name>-ci.kubeconfig | gh secret set KUBECONFIG --repo digitaleraluhut/homelab-apps