Skip to main content

​
Connect OpenClaw

If you already run OpenClaw locally or on a VPS, you can connect it to Hubify’s intelligence network without changing your existing setup. Your gateway, model routing, and local filesystem stay exactly as they are. What you gain is cloud memory backup, access to the skill registry, learning reports, and vault credentials.
hubify connect currently performs polling-based sync (every 30 seconds) with push/pull of memory files. Real-time WebSocket sync is planned for a future update.

​
Use Case

You have an OpenClaw instance running — maybe on your laptop, maybe on a VPS, maybe both. You want to:
  • Back up agent memory to Hubify’s cloud (Convex-backed, persistent)
  • Access the skill registry (5,892+ living .hub files with trust metrics)
  • Sync learnings between your local agent and your cloud workspace
  • Use vault credentials stored securely in Hubify from your local OpenClaw agent
  • See your local agent’s activity in the Hubify web dashboard
You do not need to move your OpenClaw installation or change how it runs. hubify connect sits alongside it.

​
How It Works

When you run hubify connect in a directory with a HUB.yaml, the CLI:
  1. Detects OpenClaw configuration files on your system
  2. Reads your OpenClaw user identity (user ID, username, gateway ID)
  3. Registers your local OpenClaw instance as an agent in your Hubify hub
  4. Starts syncing memory files between your local filesystem and Convex
Your OpenClaw gateway continues running as before. Hubify does not interfere with it.

​
Config File Detection

The CLI checks the following paths (in order) when looking for your OpenClaw configuration:
PathFormat
$OPENCLAW_CONFIG (env var)YAML or JSON
~/.config/openclaw/config.yamlYAML
~/.config/openclaw/openclaw.jsonJSON
~/Library/Application Support/openclaw/openclaw.jsonJSON
~/.openclaw/config.jsonJSON
If the OPENCLAW_CONFIG environment variable is set, that path is checked first. The CLI extracts your userId, username, and gatewayId from whichever config file it finds.

​
Steps

1

Navigate to your OpenClaw workspace

Go to the directory where your OpenClaw workspace lives — this is typically where your agent files, memory directory, and configuration are stored.
cd ~/clawd
2

Create HUB.yaml

If you do not already have a HUB.yaml, create one with:
hubify init
This creates a HUB.yaml manifest that links your local workspace to a Hubify hub. The CLI will prompt you for a hub name and subdomain.Alternatively, create one manually:
hub:
  version: "2.0"
  id: "hub_your_id_here"
  name: "My OpenClaw Workspace"
  owner: "yourusername"
  template: "minimal"

intelligence:
  global: opt-in

memory:
  hub_id: "hub_your_id_here"
  retention_days: 365

context:
  reserved_files:
    - AGENTS.md
    - SOUL.md
    - USER.md
    - MEMORY.md
    - HEARTBEAT.md
  memory_dir: "./memory"
  learnings_dir: "./learnings"

skills:
  registry: "hubify"
  installed: []

mode: "open"
3

Connect to Hubify

Run the connect command:
hubify connect
The CLI auto-detects your OpenClaw installation (by checking for ~/.config/openclaw), reads your user identity, and registers your local agent with your Hubify hub.To keep sync running continuously:
hubify connect --watch
4

Verify the connection

Check that your local agent appears in the hub:
hubify status
You should see your local OpenClaw agent listed alongside any cloud agents connected to the same hub.

​
What You Gain

FeatureDetails
Cloud memory backupLocal memory files (daily logs, intel, learnings) are pushed to Convex and persisted
Skill registry accessInstall from 5,892+ skills with hubify install <skill>
Learning reportsExecution reports contribute to the collective learning system
Vault credentialsAccess encrypted API keys and service credentials stored in your Hubify vault
Dashboard visibilityYour local agent’s activity appears in the web dashboard at yourname.hubify.com

​
What Stays Local

ComponentDetails
OpenClaw gatewayYour gateway process, port, and configuration are untouched
Model routingYour local model routing configuration stays as-is
Local filesystemYour codebase, project files, and shell environment are not synced
Platform configs.cursor/, .vscode/, and other IDE configs remain local

​
Running on a VPS

The process is identical for a VPS-hosted OpenClaw instance. SSH into your VPS, install the Hubify CLI, and run the same commands: 
# On your VPS
npm install -g hubify    # or: bun install -g hubify
hubify auth login
cd /path/to/your/openclaw/workspace
hubify init              # if no HUB.yaml exists
hubify connect --watch
For VPS deployments, run hubify connect --watch inside a tmux or screen session so it persists after you disconnect from SSH.

​
Multiple OpenClaw Instances

You can connect multiple OpenClaw instances to the same hub. Each registers as a separate agent with its own machine ID (derived from hostname + platform). Memory from all connected agents flows into the same Convex hub. 
Hub: Houston's OS
  Agent 1: macbook-air (openclaw, local)
  Agent 2: research-vps (openclaw, vps)
  Agent 3: hubify-cloud (cloud workspace)
All three agents read and write to the same memory store. Learnings from the VPS agent are visible to your local agent and your cloud workspace.

​
Next Steps

Local Sync

Deep dive into sync mechanics and troubleshooting

Cloud Workspace

What your cloud workspace provides

OpenClaw Integration

Full OpenClaw integration guide

Skills Registry

Browse and install from 5,892+ skills