Self-service AI workspace platform for launching Claude Code environments via Kubernetes.
https://github.com/davidbmar/everyone-ai · private · shipped
A portal that allows team members to launch pre-configured VS Code workspaces backed by Claude Code without local setup. It manages Kubernetes pods, ingress routes, and workspace templates through a Node.js API and static frontend.
npx afterburner bootstrap --host <ip> --ssh-key ~/.ssh/id_rsa npx afterburner install helm install everyone-ai-portal ./charts/portal -f profiles/k3s.yaml -f config.yaml --namespace myteam --create-namespace open https://portal.<your-domain>
flowchart TD
User[User VPN] --> Traefik[Traefik Ingress]
Traefik --> Portal[Portal Server Node Express]
Portal --> K8sClient[K8s Client Module]
K8sClient --> WorkspacePods[Workspace Pods code-server]
K8sClient --> TargetPods[Target Pods user apps]
Portal --> Config[Config JS Team Identity]
Built with Node.js/Express for the backend API, vanilla JavaScript for the frontend SPA, and Helm charts for Kubernetes deployment. Uses k3s for lightweight cluster management and Traefik for ingress routing. Includes a CLI tool called afterburner for bootstrapping and management.
sequenceDiagram
participant User
participant Browser
participant Portal
participant K8s
participant Workspace
User->>Browser: Click Launch
Browser->>Portal: POST /api/workspaces
Portal->>K8s: Create Pod and IngressRoute
K8s-->>Portal: Pod Created
Portal-->>Browser: Return Workspace ID
Browser->>Portal: GET /api/workspaces/id/events
Portal->>K8s: Watch Pod Status
K8s-->>Portal: Pod Ready
Portal-->>Browser: SSE Event Ready
Browser->>Workspace: Open VS Code URL
Operators use the afterburner CLI to bootstrap k3s on Linux VMs and deploy the portal via Helm. End users connect via VPN, access the portal URL, select a template from the App Catalog, and launch a workspace to get a browser-based VS Code instance with Claude Code ready.