Skip to main content

​
Prerequisites

Before you begin, ensure you have the following:

​
Installation

1

Clone the repository

Download the course repository using a shallow clone to get only the latest code:
git clone --depth 1 https://github.com/langchain-ai/lca-reliable-agents.git
cd lca-reliable-agents/ts
2

Install dependencies

Install all required packages using npm:
npm install
This will install all dependencies listed in package.json, including:
  • LangChain and LangGraph libraries
  • LangSmith for tracing and evaluation
  • TypeScript and tsx for running TypeScript files
  • All other required packages
3

Create environment file

Make a copy of the example environment file:
cp example.env .env
4

Configure API keys

Edit the .env file to include your API keys:
# Required
OPENAI_API_KEY='your_openai_api_key_here'

# Optional, used by the question generator
ANTHROPIC_API_KEY='your_anthropic_api_key_here'

# Required for evaluation and tracing
LANGSMITH_API_KEY='your_langsmith_api_key_here'
LANGSMITH_TRACING=true
LANGSMITH_PROJECT=lca-reliable-agents

# Uncomment the following if you are on the EU instance:
#LANGSMITH_ENDPOINT=https://eu.api.smith.langchain.com
Where to get API keys:
5

Verify setup

Run the verification script to ensure everything is configured correctly:
npx tsx env_utils.ts
The verification script checks:
  • ✅ Node.js executable location and version (must be >=18)
  • ✅ npm and tsx are available
  • ✅ Required packages are installed with correct versions
  • ✅ Environment variables (API keys) are properly configured
If the script flags any issues, see the troubleshooting section below.

​
Running Scripts

This course uses TypeScript scripts. Use npx tsx to run them directly without a build step: 
# Example: run the agent interactively
npx tsx officeflow-agent/agent_v5.ts

# Example: run an experiment
npx tsx module-2/lesson-3/run_experiment.ts
The tsx package allows you to run TypeScript files directly without compiling to JavaScript first. It’s included as a dev dependency and installed with npm install.

​
Node.js Version Management

Managing your Node.js version is often best done with a version manager. This allows you to select a Node.js version for the course independent of the system version. nvm (Node Version Manager) lets you install and switch between multiple Node.js versions: 
# Install Node.js 22 (or any version >= 18)
nvm install 22
nvm use 22

# Install dependencies
npm install
The LTS (Long Term Support) version of Node.js is recommended for stability. As of this writing, Node.js 22 is the current LTS version.

​
Using the official installer

Download and install Node.js directly from nodejs.org. The LTS version is recommended. After installation, verify your Node.js version: 
node --version  # Should be >= 18
npm --version

​
Environment Variables

This course uses the dotenv module to read key-value pairs from the .env file and set them in the environment. They do not need to be set globally in your system environment.
If you have API keys already set in your system environment, they may conflict with the ones in your .env file. The env_utils.ts verification script will detect and warn you about such conflicts. By default, dotenv.config() does not override existing environment variables.

​
LangSmith Configuration

LangSmith is required for evaluation and tracing in this course:
1

Create account

Sign up for a free LangSmith account
2

Generate API key

Navigate to Settings → API Keys and create a new API key
3

Update .env file

Add your LangSmith API key to the .env file:
LANGSMITH_API_KEY='your_langsmith_api_key_here'
LANGSMITH_TRACING=true
LANGSMITH_PROJECT=lca-reliable-agents
If you’re on the EU instance of LangSmith, uncomment and use LANGSMITH_ENDPOINT=https://eu.api.smith.langchain.com in your .env file.

​
Troubleshooting

​
Module not found when running scripts

If you see an error like Cannot find module, you likely haven’t installed dependencies yet. Solution: 
npm install
Make sure you’re in the correct directory (lca-reliable-agents/ts) when running this command.

​
Environment Variable Conflicts

If you see a warning about “ENVIRONMENT VARIABLE CONFLICTS DETECTED”, you have API keys set in your system environment that differ from your .env file. Solutions:
  1. Unset the conflicting system environment variables (commands provided in warning):
    • macOS/Linux: unset VARIABLE_NAME
    • Windows PowerShell: Remove-Item Env:VARIABLE_NAME
  2. Update your .env file to match your system environment

​
LangSmith Tracing Errors

If you see “LANGSMITH_TRACING is enabled but LANGSMITH_API_KEY still has the example/placeholder value”, you need to set a valid LangSmith API key in your .env file. Solution: Replace the placeholder value in .env with your actual LangSmith API key from smith.langchain.com.

​
Wrong Node.js Version

If you see a warning about Node.js version not satisfying requirements (must be >= 18): Solutions:
  • Install Node.js from nodejs.org (LTS version recommended)
  • Use nvm to install and manage Node.js versions

​
tsx command not found

If you get an error that tsx is not found: Solution: Make sure you’ve installed dependencies: 
npm install
Always use npx tsx (not just tsx) to run TypeScript files, which will use the locally installed version.

​
Model Providers

The course primarily uses OpenAI’s gpt-5-nano model, which is very inexpensive. If you don’t have an OpenAI API key, sign up here. You can optionally obtain an Anthropic API key here for the question generator.
This course has been created using particular models and model providers. You can use other providers, but you’ll need to update the API keys in the .env file and make necessary code changes. LangChain supports many chat model providers - see the full list here.

​
Development Environment

This course uses TypeScript scripts that you can edit and run in any editor, including:
  • VSCode
  • Cursor
  • Windsurf
  • Any text editor with a terminal
Most modern editors provide excellent TypeScript support with features like:
  • IntelliSense and autocomplete
  • Type checking
  • Go-to-definition
  • Inline documentation

​
Using Coding Agents

If you’re using coding agents during the course, you can provide your agent access to the LangSmith documentation. Instructions are available here.

Build docs developers (and LLMs) love

Get started for freeTalk to us