I’ve been using Claude a lot to refine code, generate checklists, and just generally help me move faster. But I kept finding myself typing the same kinds of instructions over and over again.

Eventually I thought, there’s got to be a better way to do this. Turns out, there is.

Claude lets you create your own custom slash commands—little reusable snippets you can call up with /whatever. You write the prompt once, save it as a markdown file, and now it’s available whenever you need it.

It’s a simple thing, but it's made my workflow a lot smoother.

A Couple Commands I Use Constantly

I’ve set up a couple of these that I use all the time.

One is for cleaning up code after I think the core functionality is done. It’s a prompt I originally saw from Aaron Francis, and it’s become one of my go-tos:

“Okay, I think all the functionality is correct. Can you please clean up anything that we might have left over or simplify anything that is too complicated?”

I saved that as a command called /cleanup-code. Now, instead of retyping it every time, I just run the command and let Claude take it from there.

Another one I use is for turning a rough task list into something more structured. Usually when I’m starting on a feature or project, I’ll dump some thoughts into Claude, then run this prompt:

Let’s break this down into clear user stories and tasks.

For each user story, write a short, human-readable title, followed by a list of atomic tasks required to build it. Group the tasks under each user story as a checklist.

Format everything in markdown, and write it as if it’s going into a build-checklist.md file that a developer will use to stay on track.

I saved that one as /build-checklist, and now I reach for it every time I need to plan something out in more detail.

How It Works

Claude looks for these custom commands in specific folders on your machine. All you have to do is save a .md file with your prompt in the right place:

  • For global (personal) commands that work across all projects, put them in: ~/.claude/commands/
  • For project-specific ones, use: your_project/.claude/commands/

The filename becomes your command. So a file called cleanup-code.md becomes accessible via /cleanup-code.

You could write these markdown files by hand—and I did that at first—but I wanted something quicker. So I made a little meta-command to help me add new commands directly from within Claude.

The meta-command

I called it /add-command, and it’s basically a shortcut for creating new shortcuts.

Instead of jumping into the terminal and creating a new markdown file every time, I can now just stay inside Claude and run:

/user:add-command Okay, I think all the functionality is correct. Can you please clean up anything that we might have left over or simplify anything that is too complicated?

That automatically creates a file called cleanup-code.md in ~/.claude/commands/, with the full prompt as its contents. From then on, /cleanup-code is ready to go.

Here’s what the prompt behind /user:add-command looks like:

# Add Custom Command

Create a new personal slash command and save it to ~/.claude/commands.

## Command Content
$ARGUMENTS

## Instructions
Take the provided command content and:

1. Generate a filename
2. Save it in ~/.claude/commands/
3. It’ll be available as /user:filename (drop the .md)

Small thing, but helpful

It’s a small thing, but when I’m deep in a flow, not having to context-switch just to save a prompt is really nice.

If you’re using Claude regularly and you’ve got a few prompts you find yourself typing again and again, this is worth setting up. Takes a few minutes, saves you a lot more.

And if you come up with any particularly useful slash commands of your own, I’d love to hear about them.