Typescript for OpenAI

Getting started with an OpenAI using Typescript using yarn for package management and nvm for managing node versions.

Set up the environment

nvm install lts/*
npm install -g strip-json-comments-cli
npm install -g yarn

Set up the project directory. I use ~/src/<project>

mkdir <applicationname>
node --version > .nvmrc
yarn init -y

Example package.json for the chatone application.

{
  "name": "chatone",
  "version": "0.1.0",
  "main": "index.js",
  "author": "Mark C Allen <mark@markcallen.com>",
  "license": "MIT"
}

Add typescript

yarn add -D typescript @types/node tsx rimraf

Create the tsconfig.json file

npx tsc --init --rootDir src --outDir dist \
--esModuleInterop --resolveJsonModule --lib es2021 --target es2021 \
--module nodenext --allowJs true --noImplicitAny true
cat tsconfig.json | strip-json-comments --no-whitespace | jq -r . > tsconfig.pretty.json && mv tsconfig.pretty.json tsconfig.json

tsconfig.json

{
  "compilerOptions": {
    "target": "es2021",
    "lib": [
      "es2021"
    ],
    "module": "nodenext",
    "rootDir": "src",
    "resolveJsonModule": true,
    "allowJs": true,
    "outDir": "dist",
    "esModuleInterop": true,
    "forceConsistentCasingInFileNames": true,
    "strict": true,
    "noImplicitAny": true,
    "skipLibCheck": true
  }
}

Get an OpenAI key add it to your .env file

# OpenAI API key for LangChain GPT-4 agent
OPENAI_API_KEY=sk-proj...

Add openai, langchain/core and dotenv

yarn add openai @langchain/core dotenv

Create a simple chat bot in the src/ directory.

mkdir src

import 'dotenv/config';
import { ChatOpenAI } from '@langchain/openai';
import readline from 'readline/promises';
import { stdin as input, stdout as output } from 'node:process';

const rl = readline.createInterface({ input, output });

const model = new ChatOpenAI({
  temperature: 0.7,
  openAIApiKey: process.env.OPENAI_API_KEY,
});

async function ask(question: string) {
  const res = await model.invoke(question);
  return res.content;
}

async function main() {
  console.log('🚀 LLM chat started. Type your questions below. Ctrl+C to exit.\n');

  while (true) {
    const input = await rl.question('❓ You: ');
    const response = await ask(input);
    console.log(`🧠 LLM: ${response}\n`);
  }
}

main();

Add scripts to the package.json

npm pkg set "scripts.build"="rimraf ./dist && tsc"
npm pkg set "scripts.start"="node dist/index.js"

Now build and run

yarn build
node dist/index.js

🚀 LLM chat started. Type your questions below. Ctrl+C to exit.

❓ You: who are you?
🧠 LLM: I am a language model AI created to assist with various tasks and provide information. How can I help you today?

❓ You:

To run without having to explicitly compile the typescript

npm pkg set "scripts.dev"="tsx src/index.ts"

Now run in dev mode

yarn dev

yarn run v1.22.22
$ tsx src/index.ts
🚀 LLM chat started. Type your questions below. Ctrl+C to exit.

❓ You:

Add everything to git.

git init

cat << EOF > .gitignore
.env
yarn-error.log
dist/
node_modules/
EOF

git add .
git commit -m "First checkin" -a

There we go, a working typescript chatbot that uses Openai.

Next steps are to add a Dockerfile for Typescript