Vault
The Lab · Vault · Blueprint · About 10 minutes

Your agent drives n8n.

Full command-line control of n8n from your app: 80+ commands, a catalogue of 543 nodes, and 40 skills for the everyday moves. One n8n instance, yours. Ten minutes once you have its API key.

In plain English

n8n is the tool where you wire things together: when this happens, do that. It has a visual editor, and it is fine for building one workflow at a time. This blueprint hands the whole thing to your agent instead. You paste one prompt, your agent installs a small command-line tool, connects it to your n8n, and from then on you say what you want in plain words: "list the workflows that failed this week", "back everything up before we touch it", "explain why that one broke". Your agent runs the commands. You read what comes back.

What you get

The n8n-cli command on your computer, connected to your n8n, plus 40 skills your agent can reach for: status, debug, backup, create, monitor and more.

Why it matters

The visual editor shows you one workflow. A command line makes all of them searchable, diffable, testable and debuggable from a chat with your agent.

How it works

One install, one API key you type yourself, one health check that proves the pipe. Then the skills. Read-only first; nothing changes until you say so.

Your seat

You own the n8n instance and the key. Your agent types every command. Your job is to recognise the words when it reports back, and to say yes or no.

Bring one n8n instance. Here's which one.

Perk
No n8n yet? Two ways in, and they are not the same.

Self-host it. n8n Cloud caps how many times your workflows can run each month, and bills you monthly for that cap. Self-hosted n8n has no cap: it runs on a server you control, and you pay for the server, not for the runs. One server, your n8n, your data; you keep it updated and backed up. And a server rarely stays a one-tool server: the next thing you self-host goes on it too.

Or n8n Cloud. The quickest start, nothing to look after. The cap is the trade.

Both are in the perks library: a Hostinger deal for the server, an n8n deal for the cloud. One is our referral link; the deal says so, and your price is never higher for it. Perks come with The Lab, claimed from your dashboard: no second account, no second login.

Self-host →n8n Cloud →Each opens Perks on its deal

One instance that you control. n8n Cloud or self-hosted, either is fine. If you run more than one, pick the one you would least mind your agent reading from, and start there. This is a connect blueprint, not a migration: first prove the pipe, then decide which workflows your agent gets to inspect or change.

You need one thing from it before you start: an API key. In your n8n, open the settings for the n8n API and create one. The key is the password for everything the CLI does, so it gets typed into your terminal by you, and it never goes anywhere else.

Never
Never paste the API key into the chat.

Your agent will tell you exactly where to type it. You type it there, yourself. The same goes for your instance's real URL in anything you save or share: the blueprint's files use placeholders (<YOUR_N8N_API_URL>, PASTE_YOUR_N8N_API_KEY_HERE) on purpose.

What happens in the ten minutes?

Five steps. Your agent does the typing; here is what goes by on the screen so none of it is a stranger.

  • It installs the toolkit. One Python package, n8n-toolkit. That is where the n8n-cli command comes from.
  • It asks you for the key, once. You type the URL and the key into the terminal yourself. The CLI saves them as a local profile called student and never asks again.
  • It checks the pipe. A health check, then a list of your active workflows. If those two come back, the connection is real.
  • It installs the 40 skills. Into your app's skill folder, so /n8n-cli-status and friends exist from the next session on.
  • You restart your app. Skills load at start. Open a new session and they are there.
You'll see
These are the real commands, in the order they run.
python3 -m pip install n8n-toolkit
n8n-cli config set-profile student --url "<YOUR_N8N_API_URL>" --key "PASTE_YOUR_N8N_API_KEY_HERE" --default
n8n-cli health
n8n-cli workflows list --active
n8n-cli skills install

The URL ends in /api/v1. If the health check fails, that ending is the first thing to look at.

Note
Codex works the same way.

The commands are identical. One difference: the install and the health check cross the network, so in Codex's sandbox those steps ask before they run. Say yes. That is the reviewer being careful, not a problem.

What you can ask your agent, from now on

Six asks that work on day one. Each one reaches for a skill the toolkit installed. Copy, paste, read what comes back.

Status

Run /n8n-cli-status and tell me, in plain sentences, which of my workflows are unhealthy and why.

Before anything changes

Use /n8n-cli-backup to export all my workflows to a local folder before we change anything. Tell me where the folder is.

Something broke

Use /n8n-cli-debug on the latest failed execution. Explain what went wrong like I am new, then give me one thing to try.

Before deleting anything

Use /n8n-cli-deps before I delete this credential: tell me every workflow that would break.

A new workflow, safely

Use /n8n-cli-create to draft a workflow that does the following, and leave it inactive until I approve it: [say what you want in plain words].

Test a webhook

Use /n8n-cli-smoke to make a smoke-test plan for this webhook, and run it only after I say go.

Words you now own

  • Workflow. One wiring diagram in n8n: a trigger, then steps. The thing you used to build by hand in the editor.
  • Execution. One run of a workflow. Failed executions are where the debugging starts.
  • API key. The password the CLI uses to talk to your n8n. You type it once; it lives in a local profile.
  • Profile. The saved pairing of one n8n URL and its key, on your computer. This blueprint makes one, called student.
  • Skill. A short instruction file your agent reads when you say the slash word. The toolkit installs 40 of them.
  • Node catalogue. The list of 543 building blocks n8n knows about, so your agent can name the right one when it drafts a workflow.

Before you share or save anything

Two habits the blueprint asks for, because a workflow export can carry a credential inside it without you noticing.

  • Export before you edit. A backup first, every time. It is one ask (the "Before anything changes" prompt above).
  • Inspect before you commit. Ask your agent to search any exported workflow for the words apiKey, credentials, password and secret before it goes into your workbench. Imported or generated workflows stay inactive until a person has read them.
Never
Never commit a real instance URL, a key, or a workflow export with secrets in it.

Your workbench is private, and this is still the rule. The placeholders in the blueprint's files are there so the shareable version never carries the real thing.

Fair questions

FAQ
I don't have n8n yet. Is this for me?

Not yet. This blueprint connects to an n8n you already run. n8n Cloud gets you one in a few minutes; self-hosting is a project of its own. Come back with an instance and an API key.

FAQ
Can it break my workflows?

Only if you ask it to change one, and the blueprint's habit is backup first, inactive until reviewed. Everything in the ten minutes is read-only: install, connect, health check, list, skills.

FAQ
Is there a workshop for this one?

No. Not scheduled, and it does not need one: the files are complete on their own. If a Tuesday ever walks it live, that session links from this page.