A slash command doesn’t make the agent smarter by itself. It points to something already prepared: a prompt, a skill, a project workflow, or a tool command. The value is that you don’t re-explain the same process from scratch every run.
Think of it like speed dial. You could type the full phone number every time, or save “Mom” or “Office” and tap that instead. The shortcut isn’t a new phone network, just a faster way to call the same place.
How it shows up
In Claude-style tools, you type / and the available commands pop up in the chat box. In a cli and terminal, it feels more like a named command in the agent environment. Either way the pattern is the same: short trigger, longer behavior. Project instructions can shape how commands behave inside a project, a skill defines the repeatable work, and the slash command is the handle you grab to run it. If the command runs a risky workflow, the saved instructions should still say what inputs it needs, what outputs to produce, and where a human reviews.
Why you care
The best use case is work you do often enough that retyping the prompt is wasteful, but important enough that you want the same process every time. A command that hides too much can get dangerous, because it feels easy while you forget what it’s actually doing. The point is making the right workflow easy to call at the exact moment you need it, not speed by itself.