AI Retro Console with n8n, LLM and Linux

Having breakfast while looking at your phone has the unwanted effect of a dopamine flood triggered by visual stimuli, clickbait, social media, and infinite scrolling.

Could an obsolete laptop be the solution?

Steps

• Install Lubuntu or any other light Linux distro in the notebook
• Install Pyfiglet with

$ pip install pyfiglet

• Create a free trial account at n8n.cloud (or sign up for any n8n hosting)
• Open this n8n template
• Configure the credentials for the tools you are going to use (Google Calendar, Gmail, Telegram, Gemini Flash API key, etc)
• Double click the "Console In" node and get the webhook URL

• Download consoleUpload script and edit the webhook URL
• Copy the script to your Linux distro and run with

$ python3 consoleUpload.py

If you want the retro look, install Cool Retro Term with

$ sudo apt install cool-retro-term

How does it work

Your old notebook is now an AI-powered retro terminal where you can enter natural language instructions.

For example:

• 'What are my next events for the afternoon?',
• 'Search and summarize headlines for TSLA',
• 'Search X for Maker Counterculture and send a Telegram with the results',
• 'Tell me if I have an important email in my Gmail inbox'
• 'Run a Linux command to block the IP 192.186.1.17'.

After a few seconds, you’ll receive the response in text mode. A log file (log.txt) stores all interactions, and more tools can be easily added via n8n

Logs

Demo

Tools and Linux commands

By default the following tools are included

• Date and time
• US markets information
• RSS news
• X search
• Gmail inbox
• Google Calendar
• Send Telegram messages

You can add more tools but remember to edit System Message for the AI Agent Node to let the agent know about the changes and how to use the new tools.

You are an AI assistant for XXXX. Answer short, cynical and dry using the following tools:

1. Search the news
2. Search the stock market
3. Search X
4. Obtain date and time
5. Send a Telegram message 
6. Search XXX's Gmail inbox
7. Search events in XXX's Google Calendar
8. Search Clarin newspaper headlines

You can also use your default knowledge base.

If you are asked to send a Telegram, use as the Telegram body, the information sent by XXX or the output from other tool.

If you are asked for news about a subject, retrieve up to 5 news and summarize

If you are asked to Search X, use as the search input the term provided by XXX. **IMPORTANT:** The X tool has strict rate limits. Do not make rapid, consecutive calls to this tool. If you need to make multiple Twitter requests, space them out. If you encounter an error related to 'Too Many Requests', you must wait a reasonable amount of time (e.g., 60 seconds) before trying again or inform the user about the rate limit.

If you are asked to search the calendar use the Google Calendar tool.

If you are asked to search XXX's inbox use the Gmail tool

If you are asked about a  Linux command, answer starting with $ and the Linux command requested only. Example, XXX says "Whats the linux command to list files", 
you answer:"$ ls"

About Linux commands

If you ask for a Linux command, it will not only return the command but also execute it, so be careful with your requests (e.g. avoid 'What's the Linux command to wipe the OS...')

Costs

For the LLM, I'm using the free tier of Google Gemini, so the only cost is n8n hosting. After using the free credits from n8n Cloud, I decided to host it on Railway ($5/month). However, since n8n is open source, you can host it on your own servers without paying for a service.

Final notes

My first idea to avoid dopamine in the mornings was to use an old notebook in text/terminal mode, but all my good intentions fell apart after struggling to configure Gmail with Mutt and Fetchmail, trying to integrate each API, and realizing that no modern website is designed for text mode (Lynx, etc.).

Using AI Retro Console, however, has proven effective in reducing dopamine triggers, while also being fun and interesting in terms of exploring the boundaries of AI agents

Last but not least, it revives the spirit of a time when computers were driven by curiosity and collaboration, rather than by control and manipulation.

TikTok

More information about this project in this TikTok