Skip to content

n8n Guide

n8n is a visual automation tool: you connect nodes on a canvas — trigger → fetch data → transform → save/notify — and n8n runs the workflow on a schedule or when an event happens. It lets you automate the boring parts of a data pipeline (re-fetch a report every morning, clean it, append it to a file, email the summary) without writing a web server.

Why it matters in this course

Automation is the CLO-3 skill of turning a one-off analysis into a reproducible process that runs without you. A scheduled n8n workflow is reproducibility in practice: same time, same steps, same output location, every run logged.

Core concepts

Concept Meaning
Workflow the whole automation, drawn as nodes connected by lines
Trigger node starts the run — a schedule (cron), webhook, or manual click
Action node does work — HTTP Request, spreadsheet/file, database, etc.
Data between nodes every node passes JSON items to the next node
Execution one run of the workflow (n8n keeps a log — your audit trail) 
(Schedule: every day 08:00) ─▶ (HTTP Request: Open-Meteo archive API)
        ─▶ (Code: reshape JSON to a flat table) ─▶ (Google Sheets / CSV file)
        ─▶ (optional: AI Agent node or notification)

Setup (once)

# Simplest: run n8n with npx (Node.js required)
npx n8n            # opens the editor at http://localhost:5678

n8n runs locally on your machine — your data and credentials stay with you (only the workflow definitions and app are downloaded). For class work, install it, create a local account, and you are ready.

A first workflow (~10 minutes)

  1. Add a Schedule Trigger node → set to every day (or every hour for testing).
  2. Add an HTTP Request node: GET https://archive-api.open-meteo.com/v1/archive with parameters for your city and date range — a keyless public API, so no account needed.
  3. Add a Code node that takes the nested JSON and returns one item per day (flatten + convert to a table).
  4. Add an output node (save to CSV/Google Sheets, or a Telegram/email node) — done: you now have an automated data collector.

The workflow-in-Python pattern (no n8n needed)

You do not need n8n installed to learn its logic. The same workflow written as Python is what Notebook 19 teaches — a plain script that fetches, transforms, and saves, wrapped so it can be scheduled:

# collect_weather.py — run it from a scheduler (cron / Task Scheduler)
import pandas as pd
import requests

def run():
    url = "https://archive-api.open-meteo.com/v1/archive"
    params = {"latitude": 31.5, "longitude": 74.3,
              "start_date": "2026-01-01", "end_date": "2026-01-07",
              "daily": "temperature_2m_max", "timezone": "Asia/Karachi"}
    data = requests.get(url, params=params, timeout=20).json()
    frame = pd.DataFrame({
        "date": data["daily"]["time"],
        "tmax": data["daily"]["temperature_2m_max"],
    })
    frame.to_csv("weather.csv", index=False)
    print(f"Saved {len(frame)} days")

if __name__ == "__main__":
    run()

Common errors

Error Fix
Workflow runs but saves nothing inspect the execution data between nodes (click the node) — the output shape is probably nested
API call fails check the URL/params in the HTTP node; many APIs need exact date formats
"Credentials required" prefer keyless APIs (Open-Meteo) for class work; never commit real credentials
Schedule never fires local n8n must be running; for always-on automation you would deploy n8n to a server
JSON too nested use a Code node to flatten before saving

Responsible automation

  • A scheduled workflow touches real files/services — keep it read-only where possible and test each node manually before enabling the schedule.
  • Log runs (n8n execution history does this automatically) so you can tell what happened and when.
  • Don't automate sending messages/emails to people without consent, and never paste private data into a cloud LLM inside a workflow.
  • Disclose automation in your README/report, exactly like any other AI-assisted step.

Go deeper