Lesson 1: Github - EDI - LLM
Chat Bot
Vibe Coding Lesson Overview
Learn how to create functional software quickly using natural language prompts and AI-powered tools.
Quick Introduction (5 minutes)
1
Explain Vibe Coding
It's the practice of describing what you want in natural language and letting an AI generate the code.
Emphasize: Anyone can create functional software or prototypes quickly.
2
State the Goal for This Hour
Objective: Understand vibe coding basics, pick a tool, build a simple app, and see how MCP servers can expand AI's capabilities.
Checklist
  • Briefly define "vibe coding" and why it's useful.
  • Clarify the hour's main outcome (e.g., build a to-do app, choose a couple of MCPs to explore).
Tool Selection & Setup (10 minutes)
Pick an IDE/Platform
Replit (free tier), bolt.new (browser-based), or Cursor (desktop).
Decide if You Need Extra AI Access
E.g., ChatGPT, Claude, or integrated AI (Replit Ghostwriter).
Basic Setup
Create/free account if needed.
Confirm environment (browser, desktop install).
Checklist
  • Choose 1 main vibe coding tool (Replit or bolt.new recommended for speed).
  • Confirm you have an AI assistant (ChatGPT, Claude, or built-in) ready.
  • Make sure everything is running (test the environment with a "Hello, world" prompt).
Build a Simple Project (15 minutes)
Ideation
Decide on a small app (e.g., to-do list, quote generator).
Prompt the AI
Provide a clear, specific description of the app's features and design.
Iterate & Test
Add or refine features (e.g., "Add checkboxes to mark tasks done").
Deploy/Share (If Time)
If using Replit or bolt.new, quickly share the URL.
Checklist
  • Spend 5 minutes describing your app in detail to the AI ("A to-do list with add/remove tasks").
  • Run the code and test at least once.
  • Iterate to add one improvement or fix (like a "clear all tasks" button).
Overview of Model Context Protocol (MCP) Servers (10 minutes)
Definition: MCP servers let AI access external tools (web scraping, debugging, game dev, etc.) safely.
Software Dev & Debugging
Gaming Development
Web Scraping & Data Extraction
Blockchain & Web3
AI Workflow & Automation
Enterprise & Business Process
Cybersecurity & Threat Detection
Creative Content & Media
Limit of 40 MCPs Active: Carefully select ones that match your needs.
Checklist
  • Briefly scan the subcategories (which domain(s) interest you?).
  • Pick 1 or 2 MCPs in your domain (e.g., GitHub MCP for code, or Fetch for web requests).
  • Note the 40 MCP limit—focus only on critical ones first.
Must-Use MCPs & Why (5 minutes)
Sequential Thinking
Step-by-step reasoning for better AI results.
GitHub MCP
Integrate with GitHub repos for reading/editing code.
Brave Search (or Web Research)
Live info beyond the AI's training data.
TaskManager
Orchestrate tasks so the AI follows a structured plan.
Knowledge Graph Memory
Extend the AI's memory beyond the current prompt.
Checklist
  • Identify one general MCP (e.g., a search or memory MCP) to experiment with if time permits.
  • Think about future expansions: "Would GitHub MCP or TaskManager help my next project?"
Best Practices & Closing (15 minutes)
1
Prompt Crafting
Be clear, specific, and iterative.
2
Testing & Debugging
After each AI-generated change, run/test. Use the AI's help to fix errors.
3
Security & Limits
Only enable MCPs you actually need. Protect credentials.
4
Next Steps
Practice building more apps with vibe coding.
Experiment with 1–2 additional MCPs for advanced scenarios (e.g., web scraping or data analysis).
Checklist
  • Run final tests on your small app to confirm it works.
  • Note any final AI prompt clarifications (e.g., "Make sure the styling is centered").
  • Evaluate whether you want to connect an MCP next time.
  • Share your project link or code snippet with classmates/colleagues if applicable.
End of Hour Outcome
1
Working Project
Have a working mini-project built via vibe coding (a simple web app or script).
2
MCP Understanding
Understand which MCP servers could enhance future tasks (with a plan to try them).
3
AI Confidence
Feel confident iterating with AI: writing prompts, refining code, and testing.
Pro Tip: Keep your final bullet points and app link as a reference. Next time, pick one new MCP to explore in-depth (like GitHub MCP or a web scraping MCP) to expand what your AI can do.
Congratulations!
Congratulations on completing your one-hour vibe coding + MCP overview session!