Python & JavaScript Environments

What are Runtime Environments?

A runtime is the engine that executes your code. Python code needs a Python interpreter. JavaScript code needs a Node.js runtime (or a browser). Before you can write a single line of agent code, these runtimes must be installed and configured correctly.

This module covers the environment management side — not the languages themselves, but the tooling around them: installing runtimes, managing packages, isolating projects from each other, and keeping everything reproducible.

Real-world usage: Every production AI system declares its exact dependencies. Anthropic's Claude SDK is a Python package installed via pip. LangChain, CrewAI, and every other agent framework are npm or pip packages. Getting the environment wrong means "works on my machine" bugs — the number one cause of wasted hours in software teams.

Key terms:

Runtime — the program that executes your code (python, node)
Package — a reusable library someone else wrote that you install into your project
Package manager — the tool that installs, updates, and removes packages (pip for Python, npm for JavaScript)
Virtual environment — an isolated bubble where packages are installed per-project, not globally

Python Setup

Installing Python:

Bash

# macOS (with Homebrew)
brew install python

# Ubuntu/Debian
sudo apt install python3 python3-pip python3-venv

# Windows
# Download from python.org or: winget install Python.Python.3.12

# Verify
python3 --version    # or just 'python' depending on your system

A common watch-out: many systems have both python (pointing to Python 2) and python3 (Python 3). Always use python3 explicitly, or set up an alias. Python 2 reached end-of-life in January 2020 — avoid it for new projects.

Virtual Environments (Python)

For any project you plan to keep or share, a virtual environment (venv) is the right call.

Without a venv, every pip install dumps packages into a single global pile. Project A needs langchain==0.3.0 and Project B needs langchain==0.2.0? Conflict. Broken imports. Hours of debugging.

A virtual environment creates an isolated directory for each project's packages. Nothing leaks between projects.

Bash

# Create a venv
python3 -m venv .venv

# Activate it
source .venv/bin/activate        # Mac/Linux
.venv\Scripts\activate           # Windows

# Your prompt changes to show you're inside the venv
(.venv) $ pip install anthropic  # Installs ONLY in this project

# Deactivate when done
deactivate

The mental model: Think of a venv as a sealed room. When activated, every pip install goes into that room. When you deactivate, you leave the room. Other projects can't see what's inside.

The .venv/ folder is local and should be in your .gitignore. It's not portable — you regenerate it from a requirements file.

pip and requirements.txt

pip is Python's package manager. It installs packages from PyPI (the Python Package Index — a public registry of 500,000+ packages).

Bash

pip install anthropic              # Install a package
pip install anthropic==0.43.0      # Install a specific version
pip install -r requirements.txt    # Install everything from a file
pip list                           # Show installed packages
pip freeze > requirements.txt      # Export current packages to a file

The requirements.txt file is how you declare your project's dependencies:

txt

anthropic==0.43.0
langchain>=0.3.0,<0.4.0
python-dotenv==1.0.1

When someone clones your project, they run pip install -r requirements.txt and get the exact same packages. This is reproducibility.

Why it matters for agents: Agent projects have deep dependency trees. The Anthropic SDK, LangChain, vector databases, web frameworks — all installed via pip. If you don't manage versions carefully, a minor update to one library can break your entire agent.

Node.js and npm

Node.js is the JavaScript runtime for server-side code. It comes bundled with npm (Node Package Manager).

Bash

# Install Node.js (which includes npm)
brew install node          # macOS
sudo apt install nodejs    # Ubuntu (consider using nvm for version management)

# Verify
node --version
npm --version

npm installs JavaScript packages from the npm registry (2 million+ packages):

Bash

npm init -y                      # Create a new project (generates package.json)
npm install axios                # Install a package
npm install -D typescript        # Install as dev dependency
npm install                      # Install everything from package.json

package.json

The package.json file is Node.js's equivalent of requirements.txt — but richer. It declares your project's dependencies, scripts, and metadata:

JSON

{
  "name": "my-agent",
  "version": "1.0.0",
  "scripts": {
    "start": "node index.js",
    "dev": "nodemon index.js",
    "build": "tsc"
  },
  "dependencies": {
    "@anthropic-ai/sdk": "^0.33.0",
    "express": "^4.18.2"
  },
  "devDependencies": {
    "typescript": "^5.3.0"
  }
}

The scripts section is powerful — npm run dev executes nodemon index.js. It's how every Node project defines its commands.

The node_modules/ folder (where npm installs packages) is massive and should always be in .gitignore. It's regenerated by running npm install.

Why it matters: Many agent tools have TypeScript/JavaScript SDKs. The Claude API has an official TypeScript SDK. MCP servers are often built in TypeScript. You need to be comfortable in both ecosystems.

Python vs. Node.js: When to Use Which

Use Case	Preferred	Why
AI/ML, data science	Python	Libraries (PyTorch, NumPy, pandas) are Python-native
Agent backends	Python	Claude SDK, LangChain, CrewAI all Python-first
Web frontends	JavaScript/TypeScript	React, Next.js, the browser
MCP servers	TypeScript	Official MCP SDK is TypeScript
API servers	Either	Python (FastAPI) or Node (Express) both work
Scripting/automation	Python	Cleaner syntax, better stdlib for file/text manipulation

In this curriculum, you'll use Python for most agent work and TypeScript for web interfaces and MCP servers. Both are essential.

Module 0.3

What are Runtime Environments?

Python Setup

Virtual Environments (Python)

pip and requirements.txt

Node.js and npm

package.json

Python vs. Node.js: When to Use Which

Sources