LangGraph Node Example

This repository showcases simple LangGraph agents written in TypeScript / JavaScript and how to run them locally or in Docker.

---

🗂️ Project structure

Path	Purpose
multi-agent.ts	Demonstrates a multi-agent conversation orchestrated by LangGraph.
simple-agent.js	Minimal working JavaScript agent.
cust-agent.ts	Custom agent example.
public/	Static assets if you decide to serve the agent via Express.
Dockerfile + docker-compose.yml	Containerised setup.

---

✋ Prerequisites

• Node.js ≥ 18 (supports fetch and ES modules)
• npm (comes with Node)
• API keys for providers you plan to use. At minimum set one of:
  • OPENAI_API_KEY
  • ANTHROPIC_API_KEY
• Optional: SERPAPI_API_KEY if you use the web-search tool.

Create a .env file at the project root with your keys:

cp .env.example .env # if the example file exists
# then edit .env and add: OPENAI_API_KEY="sk-..."

🔒 Never commit real keys to Git!

---

🚀 Quick start

# 1. Install deps
yarn install # or: npm install

# 2. Build TypeScript
npm run build

# 3. Run the multi-agent example
npm start

The start script compiles TypeScript (tsc) and executes the generated dist/multi-agent.js.

Live development

npm run dev  # incremental tsc --watch

Linting

npm run lint      # check problems
npm run lint:fix  # fix automatically

---

🐳 Running with Docker

Build and run the container (expects your .env in the same directory):

docker compose up --build

The default container command runs npm start.

---

📞 Usage example

After startup you should see chat logs similar to:

[Agent:Researcher] Searching web for: "LangGraph tutorial"
[Agent:Writer] Drafting concise answer …

Feel free to modify multi-agent.ts to experiment with different node graphs, tools or memory functions.

---

🆘 Troubleshooting

• Make sure the required API key environment variables are set.
• If TypeScript types look stale, delete dist/ and rebuild.
• On Windows PowerShell, prefix env vars when running scripts: set OPENAI_API_KEY=...; npm start.

---

📄 License

MIT