Overview
This lecture explains how to set up and use Claude Code, an AI coding assistant, including installation, best practices, integration with GitHub, AWS Bedrock, and Google Vertex AI.
Environment Setup
- Install Claude Code on Mac, Linux, or Windows (using WSL or Ubuntu via VirtualBox).
- Download and install VirtualBox and Ubuntu Desktop for Windows users.
- Allocate sufficient memory and processors to the Ubuntu VM.
- Install Node.js (version 18 or higher) and npm.
- Configure npm to store global packages locally and update your $PATH.
- Install Claude Code using
npm install -g @anthropic-ai/claude-code.
Project Initialization
- Create a project directory (e.g.,
mkdir radiocalico).
- Set up an Anthropic account, preferably starting with API access and a fixed budget.
- Claude Pro ($100/month) and Claude Max ($200/month) are options for heavy use.
- Start Claude Code by running
claude.
Best Practices for AI Coding with Claude
- Always review and test AI-generated code before accepting it.
- Ask Claude to modify small code chunks; break down complex tasks.
- Use clear, specific feedback, not vague instructions.
- Use a CLAUDE.md memory file to maintain context and preferences.
- Take advantage of command shortcuts (tab completion, up arrow, slash commands).
Example Project: Radio Calico Architecture
- The system includes a music server (HLS/Nginx), CloudFront CDN, web client (browser), Nginx web server, and Flask ratings API.
- Only the web client and backend for ratings are being built and integrated.
GitHub Integration
- Configure your GitHub credentials and set a GH_TOKEN for API access.
- Claude can commit, push, create PRs, add comments, and interact via GitHub Actions.
- Tag @claude in PRs or issues for automated reviews and suggested fixes.
AWS Bedrock Integration
- AWS Bedrock provides API access to generative AI models like Claude Sonnet and Haiku.
- Suitable for organizations needing centralized permissions and billing.
- Requires AWS account setup, model access requests, credential configuration, and model specification.
- Pricing may be higher and less transparent than direct Claude subscriptions.
Google Vertex AI Integration
- Install and configure Google Cloud SDK (
gcloud).
- Enable Vertex AI API and request model access.
- Set required environment variables and IAM configuration.
- Detailed setup instructions are available online.
Key Terms & Definitions
- Claude Code — An AI coding assistant by Anthropic for code generation and analysis.
- VirtualBox — Software for running virtual machines on your computer.
- API — Application Programming Interface; allows apps to communicate programmatically.
- GH_TOKEN — A GitHub authentication token enabling API access.
- Bedrock — AWS API platform for generative AI models.
- Vertex AI — Google Cloud platform for AI models.
Action Items / Next Steps
- Install required software (VirtualBox, Ubuntu, Node.js, npm, Claude Code).
- Set up an Anthropic and GitHub account.
- Configure environment for AWS Bedrock or Google Vertex AI as needed.
- Read the linked Anthropic documentation for further details on integrations.