This guide walks you through provisioning a Psychz cVirtual VM and installing Clawdbot, a self-hosted personal AI assistant that can bridge common chat channels (e.g., Telegram/WhatsApp/Discord) to coding/LLM agents.
1) Recommended cVirtual specs and OS
Recommended OS
Use Ubuntu 24.04 LTS (supported on cVirtual).
Recommended VM size (typical “sweet spot”)
For a reliable, always-on Clawdbot Gateway using hosted model providers (API keys), start with:
-
CPU: 2 vCPU (prefer Dedicated if you want the most consistent responsiveness; Shared is fine for cost efficiency).
-
RAM: 8 GB (this is a common starting tier on cVirtual).
-
Disk: 40+ GB (Clawdbot can run in less, but this leaves headroom for OS updates, logs, and workspace growth; Psychz supports OS images starting at 10 GB, so you can scale down if you’re cost-constrained).
If you expect heavier usage (more channels, more automations, larger workspaces), consider moving up to 4+ cores and 16 GB RAM tiers.
2) Order cVirtual
Psychz’s general flow is:
-
Log in to the Psychz Dashboard
-
Go to Services/Devices → Order Service
-
Select cVirtual and deploy the plan
If you want your article to point users directly:
-
cVirtual landing page:
https://www.psychz.net/virtual.html -
Example order page:
https://www.psychz.net/dashboard/client/web/order/plan/486
3) Create the VM (Compute → cVirtual)
After ordering:
-
In the Dashboard, open Compute → cVirtual
-
Click Create
-
Fill in:
-
Device ID (the service you just ordered)
-
Device Name
-
Image: choose Ubuntu 24.04 LTS
-
Password (12–50 characters)
-
-
Click Create and save the instance details (IP, image, vCPU, RAM, disk)
4) Connect to the VM
Option A (easy): Browser console
Psychz provides a web-based console under the VM details; you can “Create Console” and manage the VM from your browser.
Option B: SSH (standard)
From a Linux/macOS terminal:
ssh username@server_ip_address
(Replace username with root or your user, and use your VM’s public IPv4.)
5) Prepare Ubuntu (recommended baseline hardening)
Once logged in:
sudo apt update
sudo apt -y upgrade
sudo apt -y install curl ca-certificates
Security best practices Psychz recommends include enabling a firewall and using SSH key-based auth instead of passwords.
At minimum, allow SSH and enable UFW:
sudo ufw allow OpenSSH
sudo ufw enable
sudo ufw status
6) Install Clawdbot
Clawdbot’s recommended install method is the installer script (it ensures Node.js 22+ and installs the CLI).
Run:
curl -fsSL https://clawd.bot/install.sh | bash
Then run onboarding and install the daemon (recommended for servers):
clawdbot onboard --install-daemon
What onboarding does (high level)
The CLI wizard configures (based on your choices) gateway settings, model/provider auth, channels (Telegram/WhatsApp/etc.), and installs a background service (systemd user unit on Linux).
7) Keep Clawdbot running after logout (important on Linux)
On Linux, Clawdbot uses a systemd user service by default. User services can stop when you log out unless “lingering” is enabled. The Clawdbot docs recommend:
sudo loginctl enable-linger $USER
8) Verify the installation
Run the standard checks:
clawdbot doctor
clawdbot status
clawdbot health
If you configured a messaging channel, you can do an end-to-end test message (example from the docs):
clawdbot message send --target +15555550123 --message "Hello from Clawdbot"
9) Troubleshooting: “clawdbot not found” (PATH)
If the installer completes but your shell can’t find clawdbot, Clawdbot’s install docs recommend checking Node/npm and your PATH:
node -v
npm -v
npm prefix -g
echo "$PATH"
If needed, add global npm binaries to PATH:
export PATH="$(npm prefix -g)/bin:$PATH"
Then open a new terminal session.
10) Operational notes for production use
-
Don’t expose the Gateway publicly by default. The wizard’s QuickStart defaults to a local/loopback gateway and port 18789, which is safer for most setups.
-
If you do need remote access, prefer a VPN/tunnel approach and strict firewall allowlists.
-
Psychz cVirtual emphasizes NVMe-based storage, “no bandwidth costs,” and DDoS mitigation options designed for Layer 3/4 and Layer 7 patterns—helpful if your bot endpoints or web services become attack targets.