Glossary

The AI and building terms the learning path uses, in plain language. Look here whenever a session uses a word you don’t recognize.

agent: A specialist you hand a whole job to. It has its own instructions and works in its own context, then returns a finished result.Not a chatbot. A chatbot answers in one turn; an agent reads what it needs, follows a procedure, and hands back something done.
AI tool: The application you actually type in: Claude Desktop or Cursor. It changes your setup steps and where you click, not what you learn.Which one you pick is a setting, not a track. The learning path is the same either way.
Claude Artifact: The panel your finished work opens in, beside the conversation. The same content is written to a file in your project at the same moment. Private to your account until you share it — and sharing the link shares everything on the page.Not where the work is stored. The Artifact is a view of it; the file in your project is the thing you keep, and it is what later sessions build on.
Claude Code: Where the work in these sessions happens. In Claude Desktop it is the Code tab at the top left, beside Home; in Cursor it is an extension you install into the workspace window.Not a separate app to download. It comes with the AI tool you already have, and it is the part that can reach your project and run things in it.
clone: Copying your project from GitHub into your workspace, so you have a working copy to build in. GitHub keeps the original.Cloning does not move your project. Both copies exist afterwards, and syncing is how changes travel between them.
command (bash): A typed instruction to your workspace. Bash is the language those instructions are written in, the same thing you would type in a terminal.You never have to write one; Claude does. What matters is that Claude Code shows you the exact command and asks before running it, every time. Nothing runs in your workspace without you approving it.
GitHub: The free service your repository lives on. You sign up for your own account, and your project is saved into that account rather than into anything owned by BlueRock.Not a BlueRock product. It is the reason your project outlasts your subscription.
markdown (.md): Plain text with a simple way of marking headings and lists, saved in files whose names end in .md. Your skills, your agents, and your project’s memory are all written in it.Not a format you need a particular app to open. It is why you can read every part of what you build, change it by hand, and back all of it up to GitHub as text.
MCP (Model Context Protocol): A standard way for an agent to reach a tool or a data source it does not have built in. You will see the acronym in tool names and settings long before you need to build one.
repository (repo): A project folder kept in your own GitHub account, with a full history of every change. It is where your project is saved, separate from whatever machine you are working on, which is what keeps the work yours.Not the copy on your laptop or in your workspace. Those are working copies; the repository is the one that persists.
skill: A reusable workflow you trigger by name in your current chat. Written down once, then reused instead of re-explained.A skill runs in the conversation you are already in. If you would rather hand the job off and check back, you want an agent.
SSH: The standard secure way for one computer to connect to another. In setup it is what lets Claude, running on your own machine, do its work inside your cloud workspace.You do not need to understand it to use it. The Connector supplies the values; you copy them across.
subagent: The same thing as an agent. The word only emphasises that another agent called it.