Nod
Generated on 13 Nov 2026 from the Nod catalog page
Deploy Nod, the open source personal agent from the team behind n8n. Nod runs on your Droplet with its own Linux workspace and Chromium browser, connects to the model you choose, and asks before it acts.
What is Nod?
Nod is a self-hosted personal agent that works across Gmail, Calendar, Slack, Notion, and the web. Policy checks, Vault secrets, memory, and schedules run in the server — not only in the prompt.
- Self-hosted — Files, memory, and saved logins stay on your Droplet
- Browser + workspace — Docker containers give Nod its own computer
- Your model — DigitalOcean Serverless Inference, or OpenAI, Anthropic, Gemini, OpenRouter, and more
- HTTPS by IP — Caddy reverse proxy with shortlived Let’s Encrypt TLS
- Access token — Per-droplet
NOD_AUTH_TOKENgates the web UI
Key Features
- Always-on Nod server with PostgreSQL, browser, and workspace sandbox
- Optional DigitalOcean Serverless Inference via
MODEL_ACCESS_KEYat create time or first-login wizard - Caddy reverse proxy (ports 80/443 → Nod on
127.0.0.1:4310) - Helper scripts for start/stop/restart/status/update and custom domain TLS
- Ubuntu 24.04 LTS with UFW and fail2ban
System Requirements
This image requires a Droplet with at least 2 vCPU, 8 GB RAM, and a 160 GB disk (for example s-4vcpu-8gb). Simultaneous browser tasks need more capacity.
| Use Case | RAM | CPU | Storage |
|---|---|---|---|
| Minimum | 8 GB | 2 vCPU | 160 GB |
| Recommended | 16 GB | 4 vCPU | 160 GB or more |
Included System Components
- Ubuntu 24.04 LTS
- Nod
0.7.2(pinnednod-v0.7.2git tag; Docker self-host stack) - Docker Engine 25+ with Compose v2
- Caddy reverse proxy (ports 80/443 → localhost:4310)
- PostgreSQL 17 (inside Compose)
- UFW and fail2ban
Software Included
| Package | Version | License |
|---|---|---|
| Nod | 0.7.2 | Apache License 2.0 |
| Docker Engine | 29.9.3 | Apache License 2.0 |
| Docker Compose | 5.7.1 | Apache License 2.0 |
| Caddy | 2.11.9 | Apache License 2.0 |
| PostgreSQL | 17 | PostgreSQL License |
Creating an App using the Control Panel
Click the Deploy to DigitalOcean button to create a Droplet based on this 1-Click App. If you aren’t logged in, this link will prompt you to log in with your DigitalOcean account.
Creating an App using the API
In addition to creating a Droplet from the Nod 1-Click App using the control panel, you can also use the DigitalOcean API. As an example, to create a 4GB Nod Droplet in the SFO2 region, you can use the following curl command. You need to either save your API access token to an environment variable or substitute it in the command below.
curl -X POST -H 'Content-Type: application/json' \
-H 'Authorization: Bearer '$TOKEN'' -d \
'{"name":"choose_a_name","region":"sfo2","size":"s-2vcpu-4gb","image":"nod"}' \
"https://api.digitalocean.com/v2/droplets"Getting Started After Deploying Nod
Getting Started
1. Deploy the Droplet
- Select this 1-Click App from the DigitalOcean Marketplace
- Choose a Droplet size (8 GB RAM / 2 vCPU / 160 GB disk minimum, e.g.
s-4vcpu-8gb) - Add your SSH key
- Optionally set droplet environment variables
MODEL_ACCESS_KEYandINFERENCE_MODEL(orDO_INFERENCE_ROUTER) - Create the Droplet (first boot generates secrets and starts Nod)
2. Open Nod
- Visit
https://your-droplet-ip - Paste the access token shown in the SSH MOTD (also
NOD_AUTH_TOKENin/opt/nod/deployment/self-host/.env) - If Serverless Inference was pre-configured, finish onboarding with that model; otherwise connect a provider in Models
3. SSH (optional)
ssh root@your-droplet-ip
If no MODEL_ACCESS_KEY was set at create time, the first-login wizard can configure DigitalOcean Serverless Inference. Create keys at https://cloud.digitalocean.com/gen-ai/model-access-keys.
Check status with /opt/status-nod.sh. Review secrets only when needed:
sudo grep '^NOD_AUTH_TOKEN=' /opt/nod/deployment/self-host/.env
DigitalOcean Serverless Inference
Nod uses the OpenAI-compatible endpoint https://inference.do-ai.run/v1:
| Variable | Purpose |
|---|---|
MODEL_ACCESS_KEY |
Model access key (required for auto-config) |
INFERENCE_MODEL |
Model id from GET https://inference.do-ai.run/v1/models |
DO_INFERENCE_ROUTER |
Optional Intelligent Inference Router name → router:<name> |
sudo /opt/apply-inference-from-env.sh
sudo /opt/restart-nod.sh
# or interactive:
sudo /etc/setup_wizard.sh --force
Settings land in /opt/nod/deployment/self-host/.env as LLM_PROVIDER=compatible, OPENAI_COMPATIBLE_*, and LLM_MODEL.
Managing Nod
| Action | Command |
|---|---|
| Start | /opt/start-nod.sh or systemctl start nod |
| Stop | /opt/stop-nod.sh or systemctl stop nod |
| Restart | /opt/restart-nod.sh or systemctl restart nod |
| Status | /opt/status-nod.sh |
| Logs | /opt/nod-compose.sh logs -f |
| Update | sudo /opt/update-nod.sh |
| Domain TLS | sudo /opt/setup-nod-domain.sh |
Updates
sudo /opt/update-nod.sh # latest nod-v* tag
sudo /opt/update-nod.sh nod-v0.7.2 # pin a specific tag
Back up /opt/nod/deployment/self-host/.env before upgrading — especially NOD_VAULT_KEY. The database alone cannot recover saved credentials without that key. Official backup helpers live under /opt/nod/deployment/self-host/ (backup.sh / restore.sh).
Custom Domain
HTTPS already works on the droplet IP. For a hostname:
- Point DNS A record at the droplet
- Run
sudo /opt/setup-nod-domain.sh - Open
https://your-domainand use the same access token
Security Notes
- Nod binds to loopback; only Caddy is public on 80/443
- One server belongs to one person — the access token grants full control
- Keep workspace firewall settings from upstream Compose; they block cloud metadata from the sandbox
- Rotate or restrict SSH keys; use UFW as configured
Support & Docs
- Self-host guide: https://nod.n8n.io/docs/deploy/self-host/
- Configuration: https://nod.n8n.io/docs/deploy/configuration/
- GitHub: https://github.com/n8n-io/nod
- Nod is a research preview from n8n GmbH (Apache-2.0)