A collection of runnable examples showcasing key LangChain features using TypeScript.

1. Prerequisites

2. Setup

3. Examples

4. Configuration

• Node.js v16 or later

• npm (comes with Node.js)

• A code editor (e.g., VS Code)

1. Clone the repository

git clone <repo-url> langchain-ts-examples
cd langchain-ts-examples

2. Install dependencies

npm install

3. Environment variables

• Copy .env.example to .env

• Fill in your API keys:

OPENAI_API_KEY=your_openai_api_key
PINECONE_API_KEY=your_pinecone_api_key
PINECONE_ENVIRONMENT=your_pinecone_environment
PINECONE_INDEX=your_pinecone_index
ANTHROPIC_API_KEY=your_anthropic_api_key
TAVILY_API_KEY=your_tavily_api_key

Each example can be run with:

npx ts-node <path-to-example>/index.ts

Location: examples/simple/index.ts

• Demonstrates a basic chat call with OpenAI's ChatOpenAI.

Location: examples/structured-output/index.ts

• Uses zod schemas and StructuredOutputParser to enforce and parse structured JSON output from an LLM.

Location: examples/rag/index.ts

• Shows how to retrieve context from a Pinecone vector store and feed it to a ChatOpenAI model via a StateGraph.

Location: examples/tool-example/index.ts

• Defines a simple uppercase conversion tool and uses LangChain's agent executor for zero-shot tool execution.

Location: examples/langgraph-agent/agent.ts

• Builds an autonomous multi-step agent using LangGraph, integrating a third-party search tool via ToolNode.

• **package.json**: sets up scripts and dependencies

• **tsconfig.json**: TypeScript compiler options

• **.env.example**: template for environment variables