Skip to content

quick-start.md

Latest commit

 

History

History
290 lines (236 loc) · 11.9 KB

quick-start.md

File metadata and controls

290 lines (236 loc) · 11.9 KB

Quick Start - Get Your First AI App Running in 30 Minutes

Table of Contents

Prerequisites (Complete These First)

Step 0: System Prerequisites

Windows Users: Install WSL (Windows Subsystem for Linux)

Required for professional development environment

  • Install: Open PowerShell as Administrator and run: 
    wsl --install -d Ubuntu
  • Restart your computer when prompted
  • Setup: Follow the Ubuntu setup wizard after restart
  • Details: Visit Microsoft WSL Documentation for troubleshooting
  • Troubleshooting: If you encounter issues, search on Google AI mode or Perplexity, describe the problem you are encounter. It would give you good instructions.

All Platforms: Install Node.js and npm

Required for most coding agents and development tools

  • Download: Visit nodejs.org and install the LTS version
  • Verify: Open terminal and run: 
    node --version
npm --version
  • Note: Most coding agents (Claude Code, GitHub Copilot CLI, etc.) use npm for installation

All Platforms: Install Git

Required for cloning course projects

  • Download: Visit git-scm.com
  • Verify: Open terminal and run: 
    git --version

4 Essential Steps to Success

Step 1: Install Docker Desktop (5 minutes)

  • Mac: Download from docker.com → Install → Start
  • Windows: Download from docker.com → Install → Restart → Start
  • Linux: Follow Docker Engine installation guide at docs.docker.com
  • Verify: Open terminal and run: 
    docker --version
docker compose version

Step 2: Install AI Coding Agents (HIGHLY RECOMMENDED) (10 minutes)

High-Level Overview - Choose Your Path

This course was developed using three primary coding agents in rotation. Choose based on your needs:

Quick Start (Choose One):

The Three-Agent System:

All following session are assume to be running in WSL (Windows Subsystem for Linux) or Linux/Mac terminal.

Claude Code Installation

Best for: Beginners, system architecture, containerized development, project planning, PRD - product requirement documentation.

  1. Visit: claude.ai/code
  2. Install via npm: 
    npm install -g @anthropic-ai/claude-code
  3. Setup: Follow authentication process
  4. Verify: 
    claude --version
  5. Start Claude Code: 
    claude

You're Ready! Skip to Step 3: Clone and Run or continue for additional agents.

OpenAI Codex Installation

Best for: Quick prototyping, debugging, API integrations

  1. Install via npm: 
    npm install -g @openai/codex
  2. Setup API key: 
    export OPENAI_API_KEY=your_key_here
# Add to your shell profile (.bashrc, .zshrc)
  3. Verify: 
    openai --version
  4. Start OpenAI Codex: 
    codex

Gemini CLI Installation

Best for: Code review, performance optimization, alternative solutions

  1. Install via npm: 
    npm install -g @google/gemini-cli
  2. Get API key: Visit Google AI Studio
  3. Configure: 
    export GEMINI_API_KEY=your_key_here
  4. Verify: 
    gemini --version
  5. Start Gemini CLI: 
    gemini

Professional Setup

Install all three for maximum flexibility:

# Professional multi-agent setup
npm install -g @anthropic-ai/claude-code
npm install -g @openai/codex
npm install -g @google/gemini-cli

Why Multi-Agent Approach:

  • Different Strengths: Each agent excels in different areas
  • Redundancy: If one agent is down, others keep you productive
  • Real-World Practice: Enterprise teams use multiple AI tools

Usage Note: Detailed coding agent usage, software development architecture, principles and concepts will be covered in the November class sessions.

Additional Options (Optional)

  • GitHub Copilot CLI: npm install -g @githubnext/github-copilot-cli
  • Cursor IDE: Download from cursor.sh
  • Codeium: Visit codeium.com

Step 3: Clone and Run Your First AI App (15 minutes)

For Windows (WSL) and Linux/Mac:

# Clone the course repository
git clone https://github.com/pingwu/multi-ai-coding-agent.git
cd multi-ai-coding-agent

Start Your AI Application:

Task 🤖 AI-Powered 💻 Manual Command Line
Start Project 1 (** make sure you have all prerequisites installed and docker desktop is tarted **) "Bring up Project 1 (content generator) environment." make -C project-01-content-generator up
Fix API keys (if errors) "Create .env from .env.example in Project 1." `cd project-01-content-generator && [ -f .env ]
Stop Expense Tracker "Shut down the expense tracker project." make -C project-02-expense-tracker down

Then open: http://localhost:3000 (frontend) and http://localhost:8000 (backend)

Step 4: Experience Your AI App Working (5 minutes)

  1. Open browser to http://localhost:8000
  2. Type: "I spent $25 on lunch at McDonald's"
  3. Watch AI automatically categorize and save to CSV file
  4. SUCCESS! 🎉 You just ran your first multi-agent AI system!

What You Just Accomplished

✅ Professional development environment with Docker + WSL (Windows) ✅ AI-powered coding assistant ready for natural language development ✅ Multi-agent AI system processing real business expenses ✅ Production-ready containerized application running locally

⚡ Agent vs Commands Cheatsheet

Task 🤖 AI-Powered Natural Language (via Coding agents) 💻 Manual Command Line
Start Project 1 "Bring up Project 1 environment." make -C project-01-content-generator up
Start Project 2 "Start the expense tracker project." make -C project-02-expense-tracker up
Start Project 3 "Start the task tracker project." make -C project-03-task-tracker up
Start Project 4 "Start the Google OAuth task tracker." make -C project-04-Google-OAuth up
Start ADK Chat "Start the ADK quickstart demo." make up-adk
Stop Project 1 "Shut down Project 1." make -C project-01-content-generator down
Stop Project 4 "Shut down the Google OAuth task tracker." make -C project-04-Google-OAuth down
Stop ADK Chat "Stop the ADK quickstart." make down-adk
Setup OAuth (Proj 4) "Help me set up Google OAuth for Project 4." make -C project-04-Google-OAuth setup-oauth
Backend tests (Proj 1) "Run backend tests for Project 1." make -C project-01-content-generator test-backend
Follow logs (Proj 4) "Show Google OAuth task tracker logs." make -C project-04-Google-OAuth logs
Follow ADK logs "Show ADK chat logs." make logs-adk
Note: the application ports number are the same across many applications. Change to different port number if you want multiple project running simultaneously. Use coding agent to change it or guide you through it.

Next Steps with Claude Code

Try these natural language commands with Claude Code:

"Show me how this expense tracker works"
"Add a monthly report feature to this project"
"Help me deploy this to Google Cloud"
"Create a copy of this project for demo purposes"

Course Progression

  • Session 1: Explore and understand your working AI system
  • Session 2: Customize and extend using natural language development
  • Session 3: Deploy to production with professional patterns
  • Session 4: Build team collaboration features

Need Help?

Installation Issues:

Development Issues:

  • Commands not working? → See troubleshooting.md
  • Want to understand Docker? → See docker-essentials.md
  • Claude Code not responding? → Check authentication and internet connection

Course Support:

  • GitHub Issues: Report technical problems
  • Discord/Slack: Real-time help from instructors and peers
  • Office Hours: Weekly Q&A sessions

Optional: GitHub CLI Setup

Install GitHub CLI (Recommended for Smoother Experience)

The GitHub CLI simplifies authentication and repository management.

Installation:

  • macOS (with Homebrew): 
    brew install gh
  • Other platforms: Download from cli.github.com
  • npm alternative: 
    npm install -g @github/gh

Setup:

gh auth login
# Follow prompts to authenticate

Benefits:

  • Clone repositories without credential management
  • Create pull requests from command line
  • Seamless integration with course projects

Remember: Professional AI development starts with proper tooling. You're building enterprise-grade skills from day one! 🚀