Slash Command

A friend of mine started using Claude recently. Not technical, but wildly productive, the kind of person who gets more done before lunch than most people manage all week. I was teaching him a handoff prompt: a chunk of text you paste in to wrap up one chat and cleanly hand the work to the next, so you don't lose the thread when you clear context. Good habit.

Then I watched him save that prompt as a snippet in Paste, the clipboard app, so he could paste it again next time. Smart instinct. But I stopped him: "just make a slash command." He'd never heard of one. That little moment is the whole reason this entry exists, because once you see it, you can't unsee it.

Here's the idea. Instead of typing (or re-pasting) the same long instructions again and again, you save them behind a short name and fire the whole thing by typing a slash and that name. /review. /status. /publish. One keystroke runs a routine you set up once.

Think of "the usual" at your coffee shop. You don't recite the order. You say two words and the whole thing happens. A slash command is your usual, for any task you repeat.

The part that surprises people: a slash command is literally just a text file with your prompt in it. The filename becomes the command. Here's one of mine, the entire file, saved at ~/.claude/commands/code-review.md:

I need a code review. Please summarize in chat:

1. Problem/Requirement: What needed to be fixed or built
2. Solution: Your approach and the key decisions you made
3. Changes Made: Each modified file with a brief description
4. Testing: What you tested and how I can verify it works
5. Risks/Notes: Edge cases, dependencies, anything needing a second look

That's it. Save that as code-review.md in your commands folder and now typing /code-review runs all five steps. No code. Just your words, in a file, with a name.

And here's the thing I love: a slash command doesn't have to be technical at all. This is a real one of mine that I run when I want help marketing something. The $ARGUMENTS bit is where whatever I type after the command gets dropped in, so /help_me_market our new planner launch feeds that line straight into the prompt. The file, at ~/.claude/commands/help_me_market.md:

Channel your inner Alex Hormozi and put on the marketer hat. Help me
make this shine and actually drive revenue.

We want to talk about:

$ARGUMENTS

Think hard and suggest 3 angles, with the mental model behind each.
Where a screenshot would make it land harder, mark it like this:
<SCREENSHOT>what to show here</SCREENSHOT>

If you want to get fancy, you can add a few optional lines at the very top, fenced between --- marks. It's called frontmatter, and it just labels the command so it shows up nicely in the menu:

description: "Help me market something, Hormozi style"
argument-hint: "[what you're launching]"

But you don't need it to start. The prompt is the part that matters.

In my own commands folder, issues.md becomes /issues, pr-ready.md becomes /pr-ready, status.md becomes /status. The filename is the command. That's the whole rule.

You've already seen the consumer version of this: Slack's /giphy, or /commit and /test in AI coding tools. Same move. You're just making your own now.

It pairs tightly with skills: the skill is the saved know-how (the detailed instructions and context), the slash command is the button that runs it. Many of the best commands just point the AI at a CLAUDE.md-style set of rules so it shows up already knowing how you work.

My rule of thumb: anything you type more than twice is a candidate. Name it, slash it, never type it out again. My friend has about a dozen now. The snippets app never got opened again.