LangChain + LangGraph Playground

Build practical AI workflows with clean demos, memory, tools, and orchestration.

---

✨ Why This Repo

This project is designed as a serious learning and prototyping workspace for modern LLM application patterns in Node.js.

• ⚡ Fast demos for each core concept
• 🧠 Real retrieval pipeline with embeddings + vector search
• 🛠 Tool-calling agents for practical workflows
• 🔁 Stateful memory with LangGraph checkpoints
• 🧩 Supervisor-driven multi-agent architecture

---

🧱 Core Modules

Module	Purpose	File
LLM Basics	Minimal model invocation smoke test	src/basics.js
Prompt Chains	Prompt template + parser chaining	src/chains.js
RAG	Chunking, embeddings, retriever, grounded QA	src/rag.js
Agent + Tools	Tool-calling with calculator + KB	src/agent.js
LangGraph Memory	Stateful single-agent memory thread	langgraph/index.langgraph.js
Multi-Agent Supervisor	Routing between Messi and Math experts	langgraph/multiagent.langgraph.js

---

🗺 Architecture Snapshot

flowchart LR
	U[User Question] --> S[Supervisor]
	S --> M[Messi Expert Agent]
	S --> X[Math Expert Agent]
	M --> K[Knowledge Base Tool]
	X --> C[Calculator Tool]
	K --> V[Vector Store Retriever]
	V --> D[data.txt]
	M --> R[Final Response]
	X --> R

Loading

---

📁 Project Structure

.
├── data.txt
├── langgraph/
│   ├── index.langgraph.js
│   ├── logger.langgraph.js
│   ├── multiagent.langgraph.js
│   └── shared.langgraph.js
├── src/
│   ├── agent.js
│   ├── basics.js
│   ├── chains.js
│   └── rag.js
├── .env.example
├── .gitignore
├── package.json
└── README.md

---

🚀 Quick Start

1) Install dependencies

npm install

2) Create your environment file

cp .env.example .env

Windows PowerShell:

Copy-Item .env.example .env

3) Add your Gemini API key

GOOGLE_API_KEY=your_real_key_here

---

🎮 Run Demos

npm run demo:basics
npm run demo:chains
npm run demo:rag
npm run demo:agent
npm run demo:langgraph
npm run demo:multiagent

---

📜 Script Reference

Command	Runs
npm run demo:basics	src/basics.js
npm run demo:chains	src/chains.js
npm run demo:rag	src/rag.js
npm run demo:agent	src/agent.js
npm run demo:langgraph	langgraph/index.langgraph.js
npm run demo:multiagent	langgraph/multiagent.langgraph.js

---

🔐 Security Notes

• Keep .env private. Never commit secrets.
• Calculator tools currently use JavaScript eval for demo speed.
• Do not expose current calculator logic directly to untrusted user input.

---

🛠 Troubleshooting

• Missing API key: confirm GOOGLE_API_KEY is set in .env.
• Module errors: rerun npm install.
• Weak RAG answers: check data.txt quality and relevance.

---

📈 Professional Upgrade Path

• Replace eval with a safe expression parser.
• Add ESLint + Prettier + CI workflow.
• Add unit tests for tools, retrieval pipeline, and routing behavior.
• Extract shared model/retriever setup into reusable modules.

---

📄 License

ISC (see package.json).