Vibecoding
How to make AI write production code without spiraling into a chaotic mess. Our flagship course based on ECC.
0 What vibecoding is
What is vibecoding: how to build websites without code
Vibecoding in plain English: you explain the task in words, and the AI writes the code. What a vibecoder is, how they differ from a programmer, and how to start today.
1 Quick Fixer
AI Agent vs. Chatbot: the difference, in plain words
How an AI agent differs from a chatbot: a chatbot talks, an agent acts with its own hands - it creates files, runs code, fixes errors. Get this, and vibecoding finally clicks.
1 Quick Fixer 2 One-Click Wonder
Vibecoding tools: how a beginner writes code with AI
Chatbots, AI editors, terminal agents and site builders - which vibecoding tool should a beginner pick, and how do they actually differ. No fluff.
1 Quick Fixer 2 One-Click Wonder
1 How the agent thinks
How an AI agent works, in plain English: harness, context, and memory
How an AI agent actually works under the hood: the harness, the context window, tokens, and memory. A plain-English breakdown for vibecoders - so Claude and GPT get it right on the first try, not the tenth.
3 Docs Feeder 4 The Guide 5 Almost Architect
Tokens and the price of AI: how not to go broke using it
What tokens are, how much AI actually costs, and how to pay pennies for it: pick the depth of the answer, count your tokens, plug the budget leaks, and match the model to the job.
2 One-Click Wonder 3 Docs Feeder 4 The Guide
AI Agent Memory: How Claude Remembers You Between Sessions
How AI agent memory works: what instincts, confidence, and memory scope are, why an agent's long-term memory lives in files rather than the model, and how to set it up.
4 The Guide 5 Almost Architect
Compacting an agent's memory: when to clear context without losing what matters
Compacting an agent's memory in plain words: what gets squashed and lost, what survives, and the exact moment when clearing context is safe - and when it's shooting yourself in the foot.
3 Docs Feeder 4 The Guide
Your Agent Went Dumb: How to Find the Cause and Fix Your AI Agent
Agent stuck in a loop, going in circles, or talking nonsense? We walk through a checklist of why an AI agent goes dumb and how to fix the actual cause — calmly, like a mechanic, not a panicked driver.
4 The Guide 5 Almost Architect
Autonomous agent loops: how AI works on its own
Autonomous AI agent loops in plain English: a pipeline of steps, a bridge file for memory, a kill switch against runaway spending, and a cleanup step. Run an agent overnight with no nasty surprises.
4 The Guide 5 Almost Architect
2 How to ask the right way
How to Write a Prompt for Your AI Agent and Nail It on the First Try
Why does your AI agent do the wrong thing? Learn to write prompts with the WHAT-WHERE-DONE-DON'T formula and get the result on the first try, not the tenth.
1 Quick Fixer 2 One-Click Wonder
Search First, Write Later: The Prompt That Saves You Hours
The vibecoder habit of search first, write later: how to make your AI agent hunt for ready-made libraries and solutions before writing code — and save yourself hours, money, and headaches.
1 Quick Fixer 2 One-Click Wonder
Regex or a neural net: parsing text without torching your budget
Regex or a neural net for parsing text? Learn how to choose: cheap rules for tidy lines, a neural net for chaos, and a hybrid that slashes costs by 95%.
3 Docs Feeder 4 The Guide
Blueprints in vibecoding: from idea to plan in one prompt
A blueprint in vibecoding is a full project plan grown from a single-sentence idea. You learn to slice a task into steps, spot dependencies, and ask the agent in a way that won't leave you with a mess.
2 One-Click Wonder 3 Docs Feeder
3 Make it pretty
Design Without the AI Template: Killing That Neuro Site Look
Why AI keeps spitting out the same purple sites with giant letters, and how one brief — a design direction — gets you a design with character instead of a template.
2 One-Click Wonder 3 Docs Feeder 4 The Guide
How to Make an Interface Feel Good: the Tiny Things That Decide Everything
What actually makes an interface feel expensive: spacing, shadows, corner radii, button feedback, tap targets. And how to ask AI for all of it in plain words.
2 One-Click Wonder 3 Docs Feeder
Website Animations: The Kind That Help, Not Annoy
Why a website needs animations, which ones actually help, and which ones annoy and slow things down. Duration, smoothness, accessibility — and how to ask AI to get motion right.
3 Docs Feeder 4 The Guide
React and Next.js in plain English: components, props, state
React and Next.js explained simply for vibecoders: what components, props and state actually are, how the server differs from the browser, and which prompts get your agent to build a tidy site.
3 Docs Feeder 4 The Guide 5 Almost Architect
Liquid glass: iOS 26 glassy buttons, ordered with a prompt
Liquid glass from iOS 26 explained simply: what a glass button is actually made of, and how to ask an AI agent for these translucent buttons and panels without trial and error.
3 Docs Feeder 4 The Guide
SEO for beginners: how people will start finding you on Google
SEO in plain words: why a site without optimization never shows up on Google, how to set up your title, speed and page structure, and climb the search results.
2 One-Click Wonder 3 Docs Feeder
4 So nothing breaks
Readable code for vibecoders: names, DRY, KISS
How to ask an AI agent for readable code: descriptive names, the DRY and KISS principles, no magic numbers. Easier to read, cheaper to change, less likely to break.
2 One-Click Wonder 3 Docs Feeder
Error Handling in Code: So Nothing Fails Silently
Error handling for vibecoders: a clear message for humans, a full log for you, smart request retries, and a ready-made prompt so your agent stops hiding failures.
3 Docs Feeder 4 The Guide
Tests and TDD in Plain English
Tests and TDD in plain English: why you write the check before the code, how the red-green cycle works, and why tests catch bugs before your users ever see them.
3 Docs Feeder 4 The Guide 5 Almost Architect
Pre-launch checks: the agent verification loop
Pre-launch site checks: a 6-step verification loop — build, types, linter, tests, security, review the changes. So your site doesn't crash on the first click.
3 Docs Feeder 4 The Guide
E2E Tests in Plain English: A Robot That Tests Your Site for You
E2E tests in plain English: why a vibecoder needs a robot that pokes buttons on a site by itself, how to write E2E tests in Playwright via Claude, and how not to break anything before launch.
3 Docs Feeder 4 The Guide
Regression in vibecoding: a test on the bug so the AI stops breaking things
The AI fixes a bug, breaks it again an hour later, and round it goes. How a single regression test catches what the AI itself just can’t see. A guide for vibecoders.
3 Docs Feeder 4 The Guide
Production Audit: Is Your Project Ready for Real Users?
A production audit for vibecoders: how “works on my machine” differs from “works for everyone,” which zones blow up first, and how to make the AI hand you a clear go/no-go verdict.
3 Docs Feeder 4 The Guide 5 Almost Architect
Deploying a site: how to ship to the internet without a meltdown
Deployment in plain words: how to ship updates without crashes, hide secrets in environment variables, run a health check, and roll back in seconds.
3 Docs Feeder 4 The Guide
5 Security
Website Security for Vibecoders: 4 Holes You Need to Close
How to protect your site from getting hacked when you're not a programmer. Secrets in .env, validating input and files, server-side permissions, safe logs - plus a prompt for security-review.
3 Docs Feeder 4 The Guide 5 Almost Architect
AI Agent Security: How to Scan Its Settings for Holes
The AI agent itself can become a security hole: passwords, excessive permissions, prompt injection. I'll show you how to scan its settings with a single command and close the risks.
2 One-Click Wonder 3 Docs Feeder
Hooks for Your AI Agent: Protection Against Dangerous Commands and Leaks
Hooks and hookify rules in plain language: how to give your agent an automatic bodyguard that catches rm -rf, leaked secrets, and dangerous commands before they run.
3 Docs Feeder 4 The Guide
6 Swarms and automation
A Team of AI Agents: How to Run Several Agents at Once
A team of AI agents in plain words: how to run several agents in parallel, hand each one a role, merge their reports into a single result, and not drown in the noise.
4 The Guide 5 Almost Architect
Working in Parallel with an AI Agent: How to Speed Up Vibecoding
Running tasks in parallel with an AI agent: which chunks of code you can do all at once, what has to go strictly one at a time, and how to avoid two agents brawling over the same file.
4 The Guide 5 Almost Architect
AI Agent Swarms: How to Run Several Agents at Once
AI agent swarms: how to run several agents on one project in parallel. Git worktree, dmux panels, and the DevFleet foreman - no chaos, no fights over files.
5 Almost Architect 6 Rewriter
A Council of Four AIs: How to Make a Hard Decision
The council technique: how to make the AI argue with itself across four roles so you make a hard call deliberately, instead of grabbing the first option that comes along.
3 Docs Feeder 4 The Guide
A Personal AI OS on Claude Code: Memory and Scheduling
A personal AI OS on Claude Code: memory in files, a team of specialist agents, and scheduled tasks. How to build it from four folders with no databases and no servers.
5 Almost Architect 6 Rewriter
Eval-first: checks before code when you build with AI
Eval-first in plain words: why strong vibecoders write the checks before they call the AI. Capability and regression, the pass@k metric, 15-minute chunks.
4 The Guide 5 Almost Architect
7 Superpowers
Deep Internet Research: How AI Actually Finds Facts
Deep research: how to make AI search dozens of sources, read them in full, fact-check, and come back with a report full of links and dates.
1 Quick Fixer 2 One-Click Wonder 3 Docs Feeder
Smart Search for Your AI Agent: Exa Neural Search and Fact-Checking
Smart search for your AI agent: wire up Exa neural search and iterative retrieval so the agent checks facts, finds fresh code, and cites sources instead of fantasizing.
3 Docs Feeder 4 The Guide
AI Scraper: a bot that gathers website data for you
How to build a scraper bot that crawls sites on its own, pulls the data, lets AI sort it, and drops it into a table - for free, on Gemini Flash and GitHub Actions.
3 Docs Feeder 4 The Guide
Databases for beginners: tables, indexes, and migrations
What a database actually is, in plain English: tables, indexes, migrations, and column types. We break down the biggest traps a vibecoder hits — and how to dodge them ahead of time.
3 Docs Feeder 4 The Guide 5 Almost Architect
Generating Images, Video, and Sound with AI via fal.ai
How to generate images, video, and sound with AI through fal.ai right inside your chat with the agent. Which models to pick, how to write the prompt, and how not to burn your budget.
1 Quick Fixer 2 One-Click Wonder 3 Docs Feeder
HTML presentation in the browser: slides without PowerPoint
How to make a presentation in the browser without PowerPoint or a designer: one HTML file, animations, arrow-key navigation, nothing breaks on the projector. Prompt included.
2 One-Click Wonder 3 Docs Feeder
MCP for your agent: what it is and how to plug in tools
MCP in plain English: a single connector that gives an AI agent tools, data, and templates. Why a vibecoder should care, how to plug in a server, and how not to get burned.
4 The Guide 5 Almost Architect 6 Rewriter
Beautiful Python Code: Clear Names, Types, Errors
What beautiful Python code means for a vibecoder: clear names, type hints, honest error handling, and a Pythonic style. And why an agent fixes beautiful code with ease but breaks ugly code.
2 One-Click Wonder 3 Docs Feeder 4 The Guide
How to Write Website Copy That Sounds Human (Not Like a Robot)
How to write website copy in plain, human language: how to open a paragraph, which words are junk, how to make AI write like a person and stop inventing facts. Spelled out simply.
1 Quick Fixer 2 One-Click Wonder
A Knowledge Base for Your AI Agent: How to Give It Memory
How to build a working knowledge base for an AI agent: where to store facts, decisions, and notes so it remembers them after a restart and doesn't turn its memory into a junk pile.
3 Docs Feeder 4 The Guide
Course Outline
- Chapter 0. What vibecoding is
- Chapter 1. How the agent thinks
- Chapter 2. How to ask the right way
- Chapter 3. Make it pretty
- Chapter 4. So nothing breaks
- Chapter 5. Security
- Chapter 6. Swarms and automation
- Chapter 7. Superpowers