# What is n8n? Workflows, nodes and self-hosting explained

> n8n runs a chain of steps by itself when something happens, like a form being filled in. You build that chain, called a workflow, by connecting nodes on a canvas, each doing one job: calling an app, reshaping data or asking an AI model to describe a photo. You can let n8n Cloud run it, or host n8n yourself under its Sustainable Use License.

Mindket editors · Published 2026-10-06 · Updated 2026-11-03 · https://mindket.com/blog/what-is-n8n

## In short

- Skip any tutorial step that switches a workflow to active: that's how older versions worked.
- When an AI Agent node misbehaves, check every pass it made in that execution, not only its final reply.
- Without an Enterprise license from n8n, don't let clients build or edit workflows in your n8n.

## How an n8n workflow runs, start to finish

Take the AI product photo to video generator listed on Mindket, which turns one product photo into a finished clip in six steps.

Each step is a node. The form that starts the run is the trigger, marked with a bolt icon in the editor, and every other node is an [action](https://docs.n8n.io/integrations/builtin/node-types), such as getting or sending data. No code is needed, though the [README](https://github.com/n8n-io/n8n) says a step that does need it can take JavaScript, Python and npm packages.

Nothing runs on its own until you [publish](https://docs.n8n.io/build/understand-workflows/save-and-publish-workflows) the [workflow](https://docs.n8n.io/workflows/). Before that, each [execution](https://docs.n8n.io/build/understand-workflows/understand-executions), n8n's word for one run, is a manual test you start from the editor. Once it's published, webhook and form triggers use their production URLs, schedules run at their set times and app events start runs. Publish arrived in n8n 2.0, replacing the old active and inactive toggle.

**Six steps from a form to a finished clip**

1. Form trigger. Someone fills in the form and attaches a product photo.
2. HTTP Request. Uploads the photo to Cloudinary.
3. Google Gemini. Describes the product from the photo.
4. AI agent. Creative Director plans the shots, with a chat model and an output parser hanging off it.
5. Sub-workflow. A call to a second workflow renders the clip.
6. HTTP Request. Downloads the finished video.

[![n8n's post on X announcing the n8n 2.0.0 beta: foundational improvements to security, reliability and performance, plus a safer Publish and Save model for editing live workflows, over a Version 2.0 banner](https://mindket.com/blog/shots/x-n8n-io-n8n-2-publish.webp?v=1111ea5dce1e)](https://x.com/n8n_io/status/1998101094338015405)

n8n on X, 8 December 2025

> The X posts on this page are screenshots of X's own embed, so their times are US Pacific time.

## AI agents and MCP in n8n

Step four, Creative Director, is an [AI Agent node](https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent). Its chat model and output parser hang off it rather than the main line, which is how six steps make eight nodes. That group is what n8n calls a cluster node, with the agent as its root node and the other two as sub-nodes. A workflow can connect models from OpenAI, Anthropic and Google, [among others](https://docs.n8n.io/build/integrate-ai).

In n8n's example, one execution passes through the agent several times: once to set up, again to call a tool, and once more to read the tool's answer and reply.

Two nodes speak MCP, the [Model Context Protocol](https://mindket.com/blog/what-is-mcp) that lets an AI app reach outside systems. The MCP Client Tool node lets an agent [connect to an MCP server](https://docs.n8n.io/build/integrate-ai/mcp-servers) by its URL and credentials. The MCP Server Trigger works the other way round. It turns n8n itself into an MCP server that offers its tools and workflows to MCP clients, over Server-Sent Events or streamable HTTP rather than stdio.

[![n8n's post on X announcing native MCP support in n8n, with early access to both the MCP Client tool and MCP Trigger nodes, over a video thumbnail](https://mindket.com/blog/shots/x-n8n-io-mcp-client-trigger.webp?v=75aee2a6a5f4)](https://x.com/n8n_io/status/1910272150579138633)

n8n on X, 10 April 2025

## Can you use n8n commercially?

Yes, for your own business. Your staff can build workflows, and a consultant can build automations for clients on the consultant's own instance, as long as the clients can't create or edit them. The [license FAQ](https://docs.n8n.io/n8n-community-license/community-license/license-faq) draws the line at who builds the workflows, which rules out a product where your customers build their own. White-labeling n8n with its branding removed is out too.

The code is public on [GitHub](https://github.com/n8n-io/n8n), and anyone can read and extend it. It's generally free to use and share, but under n8n's fair-code model its authors restrict commercial use. Fair-code itself [isn't a software license](https://docs.n8n.io/n8n-community-license/community-license). The license that applies is the Sustainable Use License.

The [license file](https://raw.githubusercontent.com/n8n-io/n8n/master/LICENSE.md) sets three conditions:

- You may use or modify the software only for your own internal business purposes, or for non-commercial or personal use.
- You may distribute it or provide it to others only free of charge and for non-commercial purposes.
- Files with `.ee.` in their name, or `.ee` in their folder name, are not under this license and need an n8n Enterprise License.

## n8n Cloud or self-hosted?

Self-hosting pays off when your data has to stay on machines you run, or when you need a setting [Cloud doesn't expose](https://docs.n8n.io/choose-how-to-use-n8n). Out of the box, a self-hosted n8n runs as the Community edition. A Business or Enterprise license key switches it to that edition.

Hosted or self-hosted, n8n prices its plans by execution, and [only production executions](https://docs.n8n.io/build/understand-workflows/understand-executions) count toward the quota, so testing from the editor doesn't use it up. Docker is the quickest way to try your own, and the [hosting docs](https://docs.n8n.io/deploy/host-n8n) also cover npm and several cloud providers. The README's two quick-start commands start the free Community edition:

````bash
docker volume create n8n_data
docker run -it --rm --name n8n -p 5678:5678 -v n8n_data:/home/node/.n8n docker.n8n.io/n8nio/n8n
````

**Workflow executions per month in each n8n plan's smallest tier** Pro and Business sell bigger tiers too. One execution is one run of the whole workflow, however many steps it has. Business is self-hosted only.

- Starter, hosted by n8n: 2,500
- Pro, hosted by n8n: 10,000
- Business, self-hosted only: 40,000

Source: [n8n: pricing](https://n8n.io/pricing/) (read 2026-11-03)

**n8n Cloud or self-hosted, as n8n's own comparison has it**

| | n8n Cloud (Our pick) | Self-hosted |
| --- | --- | --- |
| Ready with no installation | ✓ Yes | ✗ No |
| Needs no technical expertise | ✓ Yes | ✗ No |
| n8n hosts, updates and scales it | ✓ Yes | ✗ No |
| You control the environment and configuration | ✗ No | ✓ Yes |
| Free to run | ◐ Partly | ✓ Yes |
| Enterprise features such as SSO and projects | Paid plan | Paid plan |
| Fine for production | ✓ Yes | ✓ Yes |

**Our pick: n8n Cloud.** n8n's own guide points anyone who wants to start right away, or has no technical expertise, to Cloud. Free to run means a trial on Cloud and the Community edition on self-hosted.

Source: [n8n Docs: Choose how to use n8n](https://docs.n8n.io/choose-how-to-use-n8n) (read 2026-11-03)

## Ready-made workflows, and how big n8n is

Mindket files [n8n workflows](https://mindket.com/automations/n8n) in a tab of their own under Automations. Each workflow page draws its nodes live in n8n's own preview, with a Full screen button. An Integrations block names each app the workflow connects and what you need from it, such as an API key.

Inside n8n that key becomes a credential, stored [encrypted in n8n's database](https://docs.n8n.io/deploy/host-n8n/keep-n8n-running/backup-and-restore), and a workflow refers to it only by name and ID. Sharing a workflow within one n8n instance lets its editors use [every credential in it](https://docs.n8n.io/build/manage-workflows/share-with-others), even ones nobody shared with them directly.

The Get it button downloads a zip with the workflow's JSON and its setup notes, ready to [import into n8n](https://mindket.com/blog/import-n8n-workflow) and publish. Mindket is an independent store and isn't affiliated with n8n.

n8n's homepage counts 500+ native integrations and its README 1500+, both the vendor's own figures.

**Counts as n8n and GitHub show them**

- **209,439** stars on the n8n repository on GitHub (Source: [GitHub: n8n-io/n8n](https://github.com/n8n-io/n8n), read 2026-11-03)
- **9,000+** workflow templates, per the repository's README (Source: [GitHub: n8n-io/n8n](https://github.com/n8n-io/n8n), read 2026-11-03)
- **500+** native integrations, per n8n's homepage (Source: [n8n: homepage](https://n8n.io/), read 2026-11-03)
- **250k+** developers in its community, per n8n's homepage (Source: [n8n: homepage](https://n8n.io/), read 2026-11-03)

> The counts here are as n8n's pages and GitHub showed them on 3 November 2026.

## Frequently asked questions

### Is n8n open source?

Not quite. Its [README](https://github.com/n8n-io/n8n) calls n8n source available and self-hostable. Until 17 March 2022 the license was Apache 2.0 with Commons Clause, and since then it has been the Sustainable Use License.

### What happens to a published workflow while I edit it?

It keeps running the published version. Your edits autosave within a few seconds as new versions, but they stay drafts until you publish again. Versions you name are kept when n8n prunes old ones.

### Can Claude Code use n8n?

Yes, once an owner or admin switches on Instance-level MCP in Settings and enables each workflow for it, one by one. Clients can then search workflows and trigger the ones you exposed. From n8n 2.13.0 they can also create and edit workflows. The [MCP server page](https://docs.n8n.io/connect/connect-to-n8n-mcp-server) names Claude Code, Codex, Cursor and Gemini CLI among its clients.

### How do I debug an n8n run that failed?

Open it from the executions list. n8n lists past executions per workflow and across all workflows, and you can copy a past run's data back into the editor to debug it.

## Sources

- [n8n Docs: Welcome](https://docs.n8n.io/) (read 2026-11-02)
- [n8n Docs: Workflows](https://docs.n8n.io/workflows/) (read 2026-11-02)
- [n8n Docs: Node types](https://docs.n8n.io/integrations/builtin/node-types) (read 2026-11-02)
- [n8n Docs: Understand executions](https://docs.n8n.io/build/understand-workflows/understand-executions) (read 2026-11-02)
- [n8n Docs: Save and publish workflows](https://docs.n8n.io/build/understand-workflows/save-and-publish-workflows) (read 2026-11-02)
- [n8n Docs: Use the command line](https://docs.n8n.io/deploy/host-n8n/configure-n8n/use-the-command-line) (read 2026-11-02)
- [n8n Docs: Create and edit credentials](https://docs.n8n.io/build/understand-workflows/create-and-edit-credentials) (read 2026-11-02)
- [n8n Docs: Back up and restore](https://docs.n8n.io/deploy/host-n8n/keep-n8n-running/backup-and-restore) (read 2026-11-02)
- [n8n Docs: Share with others](https://docs.n8n.io/build/manage-workflows/share-with-others) (read 2026-11-02)
- [n8n Docs: Integrate AI](https://docs.n8n.io/build/integrate-ai) (read 2026-11-02)
- [n8n Docs: What agents do](https://docs.n8n.io/build/integrate-ai/understand-ai-components/what-agents-do) (read 2026-11-02)
- [n8n Docs: AI Agent node](https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.agent) (read 2026-11-02)
- [n8n Docs: MCP servers](https://docs.n8n.io/build/integrate-ai/mcp-servers) (read 2026-11-02)
- [n8n Docs: MCP Server Trigger node](https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-langchain.mcptrigger) (read 2026-11-02)
- [n8n Docs: Connect to n8n MCP server](https://docs.n8n.io/connect/connect-to-n8n-mcp-server) (read 2026-11-02)
- [n8n Docs: Community license](https://docs.n8n.io/n8n-community-license/community-license) (read 2026-11-02)
- [n8n Docs: License FAQ](https://docs.n8n.io/n8n-community-license/community-license/license-faq) (read 2026-11-02)
- [n8n on GitHub: LICENSE.md](https://raw.githubusercontent.com/n8n-io/n8n/master/LICENSE.md) (read 2026-11-02)
- [n8n Docs: Choose how to use n8n](https://docs.n8n.io/choose-how-to-use-n8n) (read 2026-11-03)
- [n8n Docs: Host n8n](https://docs.n8n.io/deploy/host-n8n) (read 2026-11-02)
- [GitHub: n8n-io/n8n](https://github.com/n8n-io/n8n) (read 2026-11-03)
- [n8n: homepage](https://n8n.io/) (read 2026-11-03)
- [n8n: pricing](https://n8n.io/pricing/) (read 2026-11-03)
- [n8n on X: the n8n 2.0.0 beta and its Publish model](https://x.com/n8n_io/status/1998101094338015405) (read 2026-11-03)
- [n8n on X: native MCP Client and Trigger nodes](https://x.com/n8n_io/status/1910272150579138633) (read 2026-11-03)

[Browse n8n workflows](https://mindket.com/automations/n8n)