🤖Project: Build a Chatbot

Create a full conversational AI chatbot from scratch

Project: Build a Chatbot

Let's build a complete chatbot that maintains conversation history and has a personality.

What We're Building

A CLI chatbot that:

Maintains conversation history
Has a customizable personality via system prompt
Shows token usage
Can be stopped with "exit"

Step 1: Setup

Terminal
mkdir my-chatbot
cd my-chatbot
npm init -y
npm install @anthropic-ai/sdk

Step 2: The Code

Create chatbot.js:

JavaScript
import Anthropic from "@anthropic-ai/sdk";
import readline from "readline";

const client = new Anthropic();

const SYSTEM_PROMPT = `You are Luna, a friendly and knowledgeable AI assistant.
Personality traits:
- Warm and encouraging
- Uses simple language, avoids jargon
- Gives practical examples
- Admits when you don't know something
- Keeps responses concise (under 200 words unless asked for more)`;

const conversationHistory = [];
let totalInputTokens = 0;
let totalOutputTokens = 0;

async function chat(userMessage) {
  conversationHistory.push({
    role: "user",
    content: userMessage
  });

  const response = await client.messages.create({
    model: "claude-sonnet-4-6",
    max_tokens: 1024,
    system: SYSTEM_PROMPT,
    messages: conversationHistory
  });

  const assistantMessage = response.content[0].text;
  totalInputTokens += response.usage.input_tokens;
  totalOutputTokens += response.usage.output_tokens;

  conversationHistory.push({
    role: "assistant",
    content: assistantMessage
  });

  return assistantMessage;
}

// CLI Interface
const rl = readline.createInterface({
  input: process.stdin,
  output: process.stdout
});

console.log("\n🌙 Luna is ready! Type 'exit' to quit.\n");

function prompt() {
  rl.question("You: ", async (input) => {
    const trimmed = input.trim();

    if (trimmed.toLowerCase() === "exit") {
      console.log(`\nTotal tokens used: ${totalInputTokens} input, ${totalOutputTokens} output`);
      console.log("Goodbye! 👋");
      rl.close();
      return;
    }

    if (!trimmed) {
      prompt();
      return;
    }

    try {
      const response = await chat(trimmed);
      console.log(`\nLuna: ${response}\n`);
    } catch (error) {
      console.error("Error:", error.message);
    }

    prompt();
  });
}

prompt();

Step 3: Run It

Terminal

node chatbot.js

Step 4: Customize

Try changing the SYSTEM_PROMPT to create different personalities:

Code Tutor: "You are a patient coding tutor who teaches through examples..."
Chef: "You are a creative chef who suggests recipes based on ingredients..."
Fitness Coach: "You are an encouraging fitness coach who creates workout plans..."

Challenges to Try

Add streaming — show Luna's response word by word
Save history — persist conversations to a file
Add commands — /clear, /save, /personality
Web UI — turn this into a web app with Next.js

Next up: Build a code review tool that analyzes files.

Module 8

2/9

🗺️ How to Plan an AI Project

Project: Multi-Language Translation App 🌍

2/9