Aicodeguide
AI Code Guide is a roadmap to start coding with AI
Install / Use
/learn @automata/AicodeguideREADME
<div align="center" style="font-size: 30px"> <a href="#vibe">TL;DR Just show me how to vibe code 😎!</a> </div> <br /><br />Everything you wanted to know about using AI to help you code and/or to code for you.
Introduction
The way we interact with computers and write code for them is changing. And it's a profound change: on the tools we use, the way we code and think about software products and systems.
And it's changing super fast! New LLM models are being released every week. New tools, new editors, new AI coding and "vibe coding" practices, new protocols, MCP, A2A, SLOP, ... And it's really hard to keep track of all that. Everything is scattered in different places, websites, repos, YouTube videos, etc.
That's why we decided to write this guide. It's our humble attempt to put everything together and present you the practices and tools around AI coding or AI assisted code generation, all in one place, with no fuss, in an accessible form.
-
If you're a coder but is not using AI code assistants yet, this guide is for you: it presents the most recent tools and good practices to make the most of them to help on your daily jobs. Either having AI as your copilot or being the copilot for an AI agent.
-
If you never coded before but you're interested in this new "vibe coding" thing to build your own SaaS and other software products, this guide is definitely for you: We'll try to do my best to remove obscurity and leave you with what's required to start your journey, but being super critic about what is really important and what's "just hype".
Before we start, just a small suggestion on how to read this guide. We tried to organize it in a FAQ-ish way, so feel free to search and jump for possible answers to your questions. You'll see that every section as some "Resources" listed: we keep updating those resources and you'll find the most recent ones in the top of the list.
Like I said, AI changes a lot and in a daily basis, we try our best to keep this guide updated, but if you find anything missing, please feel free to open a PR or an issue, or even jump in your discord and share your new findings with us so we can include it here!
Cool, let's start!
📚 Resources:
- Zen of AI coding by Yoav Aviram
- How I use AI to code as a software engineer by Vilson Vieira
- What I learned using AI to code for a year by Vilson Vieira
- Software Survival 3.0 by Steve Yegge
- The death of the stubborn developer by Steve Yegge
- Software Is Changing (Again) by Andrej Karpathy
- Raising an Agent podcast by Amp team
- Become an AI-augmented engineer by Mary Rose Cook
- The 70% problem: Hard truths about AI-assisted coding by Addy Osmani
- Using LLMs for code by Simon Willison
- How to Build an Agent by Thorsten Ball
- Dear Student: Yes, AI is here, you're screwed unless you take action... by Geoffrey Huntley
- The revenge of the junior developer by Steve Yegge
- How to prepare for the future of development and AI by Santiago
- The End of Programming as We Know It by Tim O'Reilly
AI coding? Vibe coding? Agentic coding?
All those terms are pretty similar. But basically AI coding is about using AI models (specially LLMs these days) and all the tooling around it to help you write software. It's also called "AI for code generation" or "code gen" for short, and there's an entire fascinating field of research and engineering, that dates back to 1950's when we used Lisp to generate code. Now we have LLMs as main engines to power code generation, and there's also some threads on neurosymbolic hybrid approaches starting to show up. AI coding is also a practice: if you're using Cursor and tab-tab-tab your way to get completions, you're "AI coding"; if you're full on using Codex's agent mode, you're also "AI coding". In summary: it's any way to use AI models to help you generate code. Generally you have people who already know how to code in this group.
Vibe coding is AI coding cranked up :-) Here you don't care much about the code being generated, you just give a prompt and expect the AI to code everything for you. The term was coined by Karpathy in 2025 and it's getting pretty popular. IMHO it's helping to democratize coding for everyone that never thought about coding before!
And then we have a new breed surfacing: agentic coding. That's when you run an agent for many rounds, by itself, in a loop, ideally with some feedback signal (tests, etc). You can either run one agent or use orchestrators like GasTown to run many agents at same time, with minor to zero human interaction.
So, in summary, no matter if you're using AI to discuss your software ideas or to help on coding only parts of your already existing code base, or if you're full on vibe coding or if you let one or 100 agents running 24/7 without intervention, you're using AI to help you generate code. Let's call it AI coding and move on.
How can I use it?
You can use AI coding in many different ways, but in summary:
-
AI is your copilot: you use AI models to augment yourself, to boost your productivity. Either by firing up ChatGPT to help you on brainstroming ideas for your SaaS; or using Cursor to autocomplete your docstrings. There are many gains here, specially for creative exploration and to automate boring parts of your work.
-
AI is the pilot: here you are the copilot. This is where the "vibe coding" happens. You turn on the Cursor Agent YOLO mode or run Claude Code with
--dangerously-skip-permissionsflag set and trust everything the agents do to generate your code. Really powerful way to automate yourself, but demands some good practices on how to design systems, tame the agents and jumping in on a spaghetti of code you actually don't know about, specially to solve errors.
You should learn and practice both!
But lean more towards copiloting and away from pure YOLO vibecoding as project complexity scales up. The more likely it is that another human (or yourself in six months) will have to maintain the code, the more important this is.
🗺️ The Roadmap
How I start?
-
If you don't know how to code and want to play with it, we recommend starting with some web-based tool like Bolt, Replit, v0, Suuper or Lovable.
-
If you already know how to code, install Claude Code, Codex, Cursor, Amp or Windsurf. You can start with the free plan and then upgrade to $20 monthly plan. Claude Code Pro plan, Codex and Cursor are both pretty good and cheap, given you'll have tons of tokens to use on most recent LLM models out there. VSCode has its own Agent Mode as well. It pairs with Github Copilot and uses and Agentic Workflow to make changes and edit files.
-
If you want a more open source alternative, try Pi (recommended), OpenCode or OpenHands.
Suggestion: We really recommend creating an account in OpenRouter. It's really easy and you'll get access to the most updated LLM models and even free versions of it.
Important note: Using Claude Code through their API/SDK is super expensive! You can easily burn hundreds of dollars a day without noticing. That's why it's recommended to start with Claude Code Pro or Max plans, so you don't have to worry about it. If you really need to use their API/SDK (either to embed it in your app or other use case), make sure to keep an eye on your usage, displayed on Anthropic's dashboard.
📚 Resources:
How I prompt for coding? AKA How I vibe code?<a id='vibe'></a>
Before mid or end-2025, it was really common to LLMs hallucinate and enter in endless cycles to fix errors. Nowadays the models are really good, but it's still interesting to follow basic principles when using them to code:
- Do not ask everything in one prompt. Only prompting "hey, build me an app for my pet store" doesn't help a software engineer and much less an AI :-) Understand your project, brainstorm first with an LLM, create a PRD (Product Requirements Document), make a plan and split them in tasks. You'll find bellow a recipe on how to use ChatGPT to create one for you.
- Give it deta
Security Score
Audited on Apr 2, 2026
