Getting started with your hosted n8n

n8n is a visual workflow automation tool — connect apps, APIs and AI models by dragging nodes onto a canvas instead of writing glue code. Your ShadowNode instance is fully hosted and ready to go. This guide takes you from first login to your first live automation.

What n8n is (and what it is not)

Think of n8n as a pipeline builder. A trigger starts a workflow (a schedule fires, a webhook is called, a new row appears), then a chain of nodes does the work — fetch data from an API, transform it, decide with an if, send it somewhere. There are 400+ built-in integrations, plus a generic HTTP node that talks to anything with an API.

Your instance is the self-hosted Community edition — the full automation engine, running on your own private container. It is not an LLM by itself: AI nodes call out to model providers (OpenAI, Anthropic, OpenRouter, …) over their APIs, so you bring your own API key. That is exactly why it needs no GPU and stays cheap.

Log in & secure your account — do this first

1. Open your instance. Use the Login URL from your ShadowNode order — it looks like yourname-apps.shadownodehosting.duckdns.org. It is served over HTTPS automatically.
2. Sign in with the username and password shown once on your order page (and in your console).
3. Set your owner account. On first run n8n asks you to create the owner — enter your own email and a strong, unique password. This binds the instance to you.

The editor in 60 seconds

• Canvas. The big grid where you place and wire nodes left-to-right.
• Node panel. The + button (top-right, or the dot on a node) opens the searchable list of triggers and actions.
• Execute / Test. Runs the workflow once right now so you can see the data flow through each node.
• Active toggle. Top-right switch — flip it on and the workflow runs on its own from then on.
• Executions. The history of every run, with the exact data at each step — your main debugging tool.

Build your first workflow

A quick one that proves everything works — fetch a fact on a schedule:

1. New Workflow. Top-left → + New Workflow. Give it a name.
2. Add a trigger. Click + → search Schedule Trigger → set it to e.g. every hour. (Or pick Manual Trigger while testing.)
3. Add an action. From the trigger, click the + → add an HTTP Request node. Method GET, URL https://uselessfacts.jsph.pl/random.json?language=en.
4. Run it. Click Test workflow. Open the HTTP node and you'll see the JSON it pulled back.
5. Do something with it. Add another node — a Set / Edit Fields node to reshape the data, or any integration to send it on. Reference earlier data with expressions like {{ $json.text }}.

That is the whole pattern: trigger → fetch → transform → deliver. Everything else is the same shape with bigger pieces.

Trigger from the outside: webhooks

To let another app or service kick off a workflow, use a Webhook trigger node. n8n gives it a public URL on your own domain. Your instance already has WEBHOOK_URL configured to your real address, so the URLs it generates point at yourname-apps.shadownodehosting.duckdns.org and work straight away — no extra setup.

• Add a Webhook node, copy its Test URL, and POST to it (curl, Postman, another app) to see the data arrive.
• Once the workflow is Active, use the Production URL — that is the one you give to external services.
• Protect it: set authentication on the node (header / basic auth) so only your caller can trigger it.

Connect AI to your workflows

This is where hosted n8n shines. Drop an AI Agent or model node into a workflow, add your provider credential (your own OpenAI / Anthropic / OpenRouter API key), and chain it with the rest: summarise an incoming email, classify a support ticket, draft a reply, extract fields from a PDF.

• Credentials are stored encrypted on your instance and reused across workflows — add the key once.
• The model runs on the provider's servers; n8n just orchestrates the calls. That is why no GPU is needed here.
• Start small: Manual Trigger → AI node → Set node, test, then wire it into a real trigger.

Activate, schedule & executions

• Activate. Flip the Active toggle so schedule and webhook triggers fire on their own. Manual triggers only run when you click.
• Watch it. The Executions tab logs every run with full input/output per node — your first stop when something misbehaves.
• Your data persists. Workflows, credentials and execution history live in your container's volume, so they survive restarts.

Good to know

• Sending email? Direct SMTP (ports 25/465/587) is blocked for abuse protection. To send mail, use an email API over HTTPS instead — SendGrid, Resend, Mailgun, Postmark — via their node or the HTTP Request node.
• Resources. Your plan has a fixed RAM/CPU cap. Heavy, parallel or huge-payload workflows may need the Pro or Power plan — idle automations are very light.
• Keep it lean. Long-running instances accumulate execution history; prune old executions occasionally to save disk.
• Private by default. It is your own isolated instance — no shared tenancy, and it can't reach our internal network.