Heidelberg AICurriculum

FULL Claude Code Tutorial For Beginners in 2026! (Become a PRO)

AI Foundations ·2026-06-30 ·2 min read

Summary written by us from the video's transcript. The video, and everything in it, is AI Foundations's work.

Learn how to set up and use Claude Code—from installation and folder organization to prompting techniques, slash commands, sub‑agents, and creating local artifacts—so you can automate tasks without writing code.

Takeaways

  • Claude Code requires the Pro plan and Git; install both before starting.
  • Prompt for the final outcome (e.g., “Rename every file…”) to let Claude act directly.
  • Use @ tags to pull in markdown files as context for queries.
  • Slash commands like /goal and /loop enable continuous, hands‑free automation.
  • Sub‑agents allow parallel execution of tasks such as researching multiple items at once.

Getting Started: Plans, Download, and Git Installation

Claude Code is only available on the Pro plan or higher; the free plan lacks execution capabilities. After subscribing, download Claude Desktop from claude.com/d and open the **Code** tab. If a prompt appears saying Git is required, install it via the provided URL (get-sc.comdowns) using the tailored prompt for your OS (Windows, Mac, Linux). Restart Claude after installation to enable code sessions.

Creating and Trusting a Project Folder

Claude Code operates within local folders that act as projects. Use the **Open folder** button to create a new directory (e.g., *CC playground*) and click **Trust workspace** so Claude can read, write, and execute files there. All subsequent actions—creating files, moving images, generating markdown—occur inside this trusted folder.

Prompting for Outcomes Instead of Questions

When speaking to Claude Code, describe the desired result rather than asking how to do it. Example: instead of asking how to rename files, say **“Rename every file in this folder to lowercase.”** Claude will scan the folder and perform the action automatically, especially when set to auto‑mode which skips confirmation prompts.

Using @ Tags and Markdown for Context

Prefix a filename with **@** to reference its contents in a prompt (e.g., **@about my business**). Claude reads the entire file and can answer queries based on that data. Store information in markdown (*.md*) files—plain text with simple symbols for headings, bullet lists, and bold—to serve as Claude’s memory and as inputs for automations.

Slash Commands: Quick Tools for Automation

Enter commands starting with **/** to trigger special behaviors: * /btw* opens a side chat for unrelated questions; * /goal* sets a persistent objective until a condition is met (e.g., find 30 competitors); * /loop* repeats a prompt on a timed interval, useful for background tasks like email scraping. These commands keep the main workflow clean and enable continuous automation.

Sub‑Agents: Parallel Workflows

Use the keyword **sub agents** to spin up multiple Claude instances that work concurrently. For example, ask Claude to research four competitors simultaneously; each sub‑agent creates its own markdown report. You can monitor running agents via the upper‑right panel and view individual transcripts, allowing parallel content creation, folder structuring, or social media posting.

Creating Local Artifacts and Organizing an OS

Artifacts are mini web apps generated from data in your folder. Prompt Claude to build a dashboard (e.g., an element study deck) and then say **“open this for me on localhost”** to launch it locally. To keep the workspace tidy, treat each major life or business area as its own folder with two files: a *claude.md* instruction file that defines Claude’s role (chief of staff) and a context folder containing relevant knowledge. This folder‑based operating system lets Claude switch personalities based on where you work.