Discover Cursor AI and n8n with crypto charts: Boost Your Automation Now!

In this in-depth article we explain how we created an automation by using Cursor AI and n8n

Introduction

I have not been developing in traditional development tools for years, but have been heavily involved in low-code projects. All the hype around new development tools made me decide to develop something small with cursor AI and n8n. The ultimate goal is to develop a generic tool that can do analytics through Chrome within online cloud solutions. These could be websites or SaaS solutions. Some examples: Visma ERP, Superoffice and others.

Why?

I believe in Citizen Developers, especially with the current generation of tools. Most companies will do just fine with simple solutions that simplify or automate work. With AI tools, you can quickly build meaningful solutions without a large and cumbersome development team.

Every SME could use an AI specialist to build internal solutions.

Furthermore, I think software integrations are going to change. You used to need APIs and structured data but now you can use AI to analyze screenshots and ask questions about them, this makes a huge difference in development costs.

Example: you are in a financial screen of your administration and want to do some analysis on sales orders. You take a screenshot via the Chrome plugin and then specify what you want to know. Voila, knowledge at your fingertips.

Which tool did I need?

I didn't really need anything but my interest in stocks and crypto gave me the idea to create a Chrome extension that does simple analysis of Tradingview.com charts. I am not a specialist but am trying to learn more and more about when to get in or not.

The tools

For these kinds of experiments you always need a backend and in this case I again chose n8n. This runs as an Open Source tool on a Digital Ocean server and provides several customers with a stable backend solution.

For the frontend, the starting point was to use an AI tool. After watching several videos, the choice for cursor AI was easily made.

The idea

My idea was simple:

A Chrome plugin that takes a screenshot and sends it to a backend service that analyzes this screenshot and then shows the analysis of a Tradingview chart in the Chrome plugin.

Indeed simple, but so should the plugin:

• Work only on the tradingview.com website;

• Have a setting for the URL you send the screenshot to;

• Do checks that everything is entered correctly;

• Be disabled if you are on the wrong website.

The details

As often happens, I just started. I already had an idea and I fleshed it out.

The Chrome plugin

I didn't have a clear interface in mind but it had to be simple:

• screenshot button

• settings

• notifications

• result

My starting point was, no complicated interface but exactly what is needed.

Now AI was around the corner through Cursor AI. After installation, the first step was to choose a folder. Late on in that folder we will automatically generate the code.

I personally had everything built through the composer via prompts, no manual tweaks.

You start with the prompt to build a Chrome plugin and work from there. It's important to describe exactly what you want, from interface to calls to external URLs. Some tips:

• Also, paste errors into the composer and ask to fix it

• Describe in detail what you want

• Also state clearly if you want assets such as icons to be generated

• Also ask to add debug code for the the Chrome console

My conclusion with Cursor AI

I spent about 2 hours working on it and it was trial and error. My first idea was to use S3 but on that path I encountered a little too many issues. This was not due to Cursor AI but an issue with the specific requirements of the S3 provider.

Overall, it was nice work and I have never made such a meaningful Chrome extension so quickly.

The backend

This was a piece of cake, we created so many workflows for clients that this was an investment of half an hour max. The main steps to accomplish this were:

• Webhook URL

• Forwarding binary file to OpenAI image analyzer

• Send response request

• This flow is super simple

OpenAI image analyzer

We do not use an advanced agent but purely an image analyzer block. Of course, you need an API key from OpenAI.

We then return the OpenAI result via the Respond to webhook block. See blow short video how the Chrome plugin works.

Basically these are the component. You can play around with it yourself, below we share the code for n8n and the Chrome plugin.