Setting Up VS Code + Claude Code
Your AI-powered development environment in 15 minutes
Get VS Code installed, configured, and connected to Claude Code so you can start building with AI assistance right away.
Step-by-step guides for setting up your development environment, learning tools, and getting started with AI-assisted coding.
Your AI-powered development environment in 15 minutes
Get VS Code installed, configured, and connected to Claude Code so you can start building with AI assistance right away.
The 10 commands you'll use every day
Learn to navigate your computer from the command line. No prior terminal experience needed — we'll start from zero.
Version control from scratch in 20 minutes
Install Git, create a GitHub account, and push your first commit. The essential workflow every developer uses daily.
Install it, configure the settings that matter, and stop at five extensions instead of fifty.
A start-to-finish VS Code setup: install, the four panels you actually use, the settings worth changing on day one, and a lean extension list that keeps the editor fast.
Get a Python you control, understand what PATH is doing, and isolate every project properly.
Your Mac already has a python3, and it is the wrong one to build on. Here is how to install Python with Homebrew or the official installer, verify it, and set up virtual environments the right way.
Tick the PATH box, learn the py launcher, and fix the two errors every Windows beginner hits.
A complete Python setup for Windows: the installer checkbox that causes most problems when missed, the py launcher, virtual environments in PowerShell, and the fixes for the two errors everyone runs into.
One mental model — four places your code can live — makes every Git command obvious.
Git stops feeling arbitrary once you know the four places your code can be. This guide teaches that model, then the daily loop, branching, honest undo commands, and how to finish a merge conflict.
Generate an ed25519 key, load it into the agent, and never paste a token into a push again.
A ten-minute setup that removes a daily annoyance: generate an ed25519 SSH key, load it into ssh-agent, register the public half with GitHub, and switch your existing repos over.
Install it, finish the PATH step everyone skips, and learn the eight commands that cover everything.
Homebrew turns hunting for .dmg downloads into one command, and upgrading everything into one more. Here is the install, the PATH step people miss, the verbs that matter, and a Brewfile that rebuilds a machine.
Pick the right version, install it with a version manager, and understand what npm just did to your folder.
Install Node and npm on any operating system, choose between LTS and Current with real version numbers, and set up your first project so the lockfile, node_modules, and global installs stop being mysterious.
Navigation, files, pipes, PATH, and the keyboard habits that make the terminal fast.
A working tour of the Unix command line on macOS and Linux: how to read any command, move around, redirect output, set your PATH, edit your shell profile, and use the shortcuts fast people rely on.
Install it, understand the three nouns, run a real container, and stop it from eating your disk.
A working Docker setup from scratch: what images, containers, and volumes actually are, installing Docker Desktop or Engine per platform, running nginx on localhost, writing a Dockerfile, and reclaiming the disk space Docker quietly consumes.
Work backwards from what you want to build, pick in an evening, then stop shopping and commit.
A decision framework instead of a popularity ranking: map the outcome you want to the language that serves it, learn which languages punish beginners, test-drive two in one evening, and commit long enough to finish something.
Install it, point it at a real repository, and build the habits that keep it useful.
Claude Code is an agentic coding tool that runs in your terminal, reads your repository, and edits files. This is the install, the first session, and the habits that separate useful from dangerous.
One mental model, four prompt ingredients, and the habits that separate useful answers from confident nonsense.
Understanding what ChatGPT actually does explains both its strengths and its failure modes. This covers model choice, prompt structure, feeding it source material, using it to learn, and knowing when to stop.
Projects for persistent context, long documents done right, and Artifacts you can actually iterate on.
Claude's real advantages are long-document work, Projects that carry context between chats, and Artifacts you can see and revise. This covers how to use each one, and where it still falls short.
Why most prompts fail, five parts that fix them, and rewrites you can copy across any model.
Most bad AI output comes from underspecified prompts, not weak models. This is the anatomy of a prompt that works, four before-and-after rewrites, and a template you can reuse anywhere.
Turn it on, learn four keys, and build the habit that keeps its suggestions honest.
A practical guide to GitHub Copilot: enabling it in VS Code, the inline suggestion loop, steering it with a signature before you type, and the failure modes nobody warns beginners about.
Three interaction modes, the context skill underneath them, and how to review an agent's work.
Cursor is a VS Code fork with AI built into the editing surface. This covers importing your setup, choosing between Tab, inline edit and the agent, feeding it context, and reviewing diffs before they land.
The process that decides whether it works: scope, spec, data model, then finally a tool.
AI builders and no-code platforms can produce a genuinely useful app in an afternoon — if you scope it honestly and get the data model right first. Here is the order that works, and where the ceiling is.
Four claim types, four checks, and a triage rule so you are not verifying everything equally.
AI models produce confident wrong answers that look exactly like right ones. Here are the specific checks for facts, numbers, citations, and code — and how to decide how hard to look.
Learn the one idea Notion is built on, then build the smallest setup that survives a real week.
Most people quit Notion because they build an elaborate system before they have anything to put in it. This walks the core ideas — pages, blocks, databases, views — then a starter setup you can finish in fifteen minutes.
One complete project from import to export, instead of a tour of panels you will never open.
A start-to-finish first edit in Premiere Pro: set up the project, match your sequence to your footage, cut with in and out points, fix audio, add a title, and export a file that does not look mushy.
Build one real screen from blank canvas to a shareable link, learning only what that requires.
A first Figma project done properly: frames, layer discipline, auto layout, a button component with variants, colour and text styles, a two-frame prototype, and a link you can send someone.
Plain Markdown files on your own disk, linked together — set up a vault and the one habit that makes it work.
Obsidian stores your notes as Markdown files in a folder you own, and treats links between notes as first-class. Here is how to set up a vault, build the linking habit, and choose a sync method honestly.
A retrieval system, not a storage system — capture rules, layered distilling, and the weekly review that holds it together.
Most personal knowledge systems become a graveyard with a good search function. This is the method side: what to capture, how to organize by purpose, how to distill in layers, and the weekly habit everything depends on.
Excel has hundreds of features and about twelve that carry almost all real work.
The Excel skills that actually get used: formula rules, the dozen essential functions, absolute references, real Tables, PivotTables, one honest chart, and the data habits that prevent pain.
Three note types, kept separate, that you will actually reread instead of a vault you never open.
Transcribing tutorial code produces notes nobody rereads. This is the note system that pays for itself: a working log, concept notes written as claims, and a snippet library, plus the review habit that keeps them alive.
Build it around the week you actually have, not the week you wish you had.
Most study schedules collapse in week two because they were designed for an idealized week. This one starts from an honest audit of your hours, anchors sessions to routines instead of clock times, and has a recovery rule built in.
What flashcards are genuinely for, how to write cards that work, and how to avoid the review avalanche.
Spaced repetition makes recall nearly free, but only for the right material and only with well-written cards. This covers the mechanism, Anki setup, the card-writing rules that decide whether it works, and the mistakes that make people quit in week three.
Triage first, match the page to the question you actually have, and always check the version.
Documentation is written as reference for people who already have the model, which is why reading it front to back fails. This covers the triage pass, the four kinds of docs, reading signatures, and what to do when the docs are thin.