Skip to main content

​
Your Cloud Workspace

A Hubify cloud workspace is a persistent, always-on AI operating system running at your personal subdomain. It is not a chatbot interface — it is a full compute environment where your agent operates 24/7 with its own filesystem, identity, memory, and skills. What a cloud workspace gives you:
  • Always-on agent — your OpenClaw gateway runs continuously, executing tasks even while you sleep
  • Web dashboard — a Next.js app served from your chosen template at your subdomain root
  • Live terminal — observe your agent in real-time through a browser-based terminal
  • Persistent storage — 10GB volume that survives machine restarts, storing memory, skills, learnings, and configuration
  • Local sync — connect your local machine with hubify connect and share memory bidirectionally

​
Accessing Your Workspace

Every workspace is available at four URLs:
URLWhat It Does
yourname.hubify.comWeb dashboard — your template’s main interface
yourname.hubify.com/chatChat interface for direct agent interaction
yourname.hubify.com/terminalLive terminal (ttyd) — observe agent activity in real-time
yourname.hubify.com/filesBrowse workspace files on the persistent volume
All routes are served over HTTPS with automatic SSL certificates.

​
Infrastructure

Each workspace runs on a dedicated Fly.io machine with the following specifications:
ResourceDetails
CPUshared-cpu-2x
RAM2GB
Storage10GB persistent volume mounted at /data
RegionAuto-selected (nearest to user)
SSLAutomatic via Caddy wildcard + Let’s Encrypt
NetworkIsolated 6PN network (user-{username}-net)
Three services run on every workspace machine: 
yourname.hubify.com
  Port 80   (nginx)           -- reverse proxy, serves dashboard
  Port 3000 (OpenClaw gateway) -- AI agent runtime, heartbeat, tool execution
  Port 8080 (ttyd)             -- live web terminal
  Port 4000 (stats server)     -- auth validation, workspace health

​
OpenClaw Gateway

The OpenClaw gateway is the core process on your workspace machine. It runs on port 3000 and handles agent sessions, heartbeat cycles, tool execution, and model routing. The gateway includes crash loop detection: if the process crashes more than 5 times within 60 seconds, the restart loop halts and the workspace enters an error state. This prevents runaway restarts from consuming resources.

​
Live Terminal

The terminal at yourname.hubify.com/terminal is powered by ttyd and provides real-time visibility into your agent’s activity. Terminal access is protected by double-layer authentication:
  1. JWT validation — nginx validates your workspace JWT before proxying the request
  2. HTTP Basic Auth — ttyd requires a username and password generated on first boot
The terminal is read-only by default. This lets you observe what your agent is doing without risking accidental interference. Write access can be enabled in workspace settings.

​
Nginx Reverse Proxy

Nginx handles all incoming traffic on port 80 and routes requests to the appropriate service:
  • / and /dashboard routes go to the Next.js dashboard
  • /terminal routes go to ttyd on port 8080
  • /api routes go to the OpenClaw gateway on port 3000
  • Sensitive file paths (.env, openclaw.json) are blocked with 403 responses

​
Dashboard

Your dashboard is a Next.js application served from your chosen template. The sections and pages vary by template:
TemplateDashboard Sections
MyOSYou, Projects, Team, Memory, Fitness
Dev OSCode, PRs, Deployments, Tech Debt
Founder OSPipeline, Content, Comms, Analytics
Research OSPapers, Synthesis, Notes, Knowledge
Client OSProjects, Client Comms, Deliverables
MinimalBlank slate with skills chooser
After deployment, the dashboard is fully customizable. Templates define a starting point, not a constraint.

​
Checking Workspace Health

You can check your workspace status from the CLI or the web dashboard. 
hubify status
The stats server on port 4000 also exposes a health endpoint that returns machine status, uptime, and resource usage.

​
Managing Workspace State

Workspaces can be in one of four states:
StateDescription
activeMachine is running, agent is live, dashboard is accessible
sleepingMachine is stopped to save resources (Free tier idle behavior)
provisioningMachine is being set up for the first time
errorMachine encountered an issue (crash loop, volume error, etc.)

​
Suspend and Resume

Free tier workspaces auto-sleep after a period of inactivity. Pro tier workspaces run 24/7. 
hubify workspace suspend
Sleeping workspaces auto-resume on any incoming request — visiting yourname.hubify.com will wake the machine. Cold start typically takes a few seconds.
Suspending a workspace does not delete any data. Your persistent volume at /data retains all files, memory, and configuration. When the machine resumes, it picks up exactly where it left off.

​
How Boot Works

When a workspace machine starts (either first deploy or resume from sleep), the boot.sh script runs:
1

Seed workspace files

If this is a fresh install (no HUB.yaml at /data), template files are copied from the Docker image to the persistent volume. User-specific values like username and hub ID are injected into USER.md and HUB.yaml.
2

Start ttyd terminal

The web terminal starts on port 8080 with read-only access and HTTP Basic Auth credentials.
3

Start dashboard

The Next.js dashboard starts on port 3001, proxied through nginx.
4

Start OpenClaw gateway

The OpenClaw gateway starts as the foreground process on port 3000 with the state directory set to /data.
5

Convex sync

The hub registers with Convex, syncs memory, and marks the workspace as active.

​
Next Steps

Local Sync

Connect your local machine to your cloud workspace

Workspace Security

Authentication, encryption, and isolation details

Subdomains

How yourname.hubify.com routing and SSL work

Workspace Modes

Guided, Power, and Open modes