Attach or run: the @ and / split

The @-attaches / -runs mental model

Two characters carry the entire daily-driver grammar of the Agent sidebar - Cursor’s chat panel, where you brief the agent on tasks - and confusing them is a common stumble. Hold one rule and you never confuse them again:

@ attaches context. / runs something.

@ pulls something into the prompt. Type @ in the Agent input and a picker opens; choose a thing and it’s attached as context the agent can see for that turn - a file or folder, a terminal, a past chat, a git diff, the browser (that’s the @ menu as of September 2026; the mentions docs track the current list). It does not do anything. It answers the question “what should the agent be looking at while it works on this?” When you say @importer.ts fix the date parsing here, the @importer.ts is you handing the agent the file so it doesn’t have to go find it.

/ triggers an action. Type / and the picker lists things the agent can run - your custom commands and skills, a skill being a packaged procedure you or your team defined (for example /write-tests, for a command you’ve defined yourself; the commands and skills docs track the current list, as of September 2026). / answers “what procedure do I want to fire?” not “what should it read?”

The test that settles nearly every case: am I telling the agent what to look at, or what to do? “Look at” is @. “Do” is / - unless it’s an instruction you haven’t packaged as a command, which is just typed as prose. A prompt routinely uses both - @money.ts @money.test.ts /add-test cover the negative-balance case attaches two files with @ and fires a custom command with / - and once the rule is in your hands you stop thinking about it. We don’t catalogue every @ target or every / command here; the full @ inventory lives in the context chapter’s @ mention reference, and defining your own / commands is an extending-chapter lesson. What you need today is the split, because it’s the grammar under every prompt you’ll type for the rest of the course.

So you can now aim the agent precisely. What happens after it answers is the half of the loop that keeps you safe. Next: reviewing what it proposes.