☁️ Run 24/7 on a VPS

⚡ The whole thing, in short

From provisioning to the first scored jobs is about 15–30 minutes, mostly SSH-key setup. The full sequence on the server:

# After you SSH into the VPS as root:
curl -fsSL https://jobhunterteam.ai/install.sh | bash   # ~1 min
exec bash -l                                            # pick up the jht PATH
jht up                                                  # pull image + start
jht setup                                               # wizard: provider + login
jht providers update                                    # install the provider CLI
jht oauth-login                                         # provider OAuth device flow
jht cloud login                                         # link the VPS to your web account
jht team start                                          # start the agents
jht team status                                         # verify

🪜 Step by step

1. Provision the server

On your cloud console (e.g. Hetzner), create an Ubuntu 24.04 box, ~4 GB RAM, in an EU region for GDPR. Upload a dedicatedSSH key — don't reuse your personal one:

ssh-keygen -t ed25519 -f ~/.ssh/jht_hetzner -C "jht-vps"
cat ~/.ssh/jht_hetzner.pub   # paste this into the cloud console

Note the server's IPv4 address when it's ready.

2. SSH in

ssh -i ~/.ssh/jht_hetzner root@<VPS_IP>

Tip: before the first connection, compare the host fingerprint shown in your console against ssh-keyscan to rule out a man-in-the-middle.

3. Add a little swap (recommended)

Eight agents can spike RAM. 2 GB of preventive swap avoids an OOM kill:

fallocate -l 2G /swapfile && chmod 600 /swapfile
mkswap /swapfile && swapon /swapfile
echo '/swapfile none swap sw 0 0' >> /etc/fstab

4. Install, configure, start

Run the one-liner and the commands from the short version above. Two things matter on a VPS specifically:

• jht cloud login— pairs the server to your web account so you can see results from any browser. On a VPS this is essential: you won't be at the terminal all day, the web dashboard is your main window into the team.
• The container runs restart: unless-stopped, so the team survives reboots automatically.