The Complete Claude Code Slash Command Atlas (91 Commands, Beginner-Friendly)
I had Claude Code compile every slash command in the latest build as of today into an atlas a beginner can follow: 91 commands, each explained in plain language by the headache it actually saves you. It's hard to find a version this detailed anywhere; worth sharing and saving.
This episode is short, and it mainly makes one point: don't stress about memorizing all 91. Plenty of them—the context-related ones, /goal for setting an objective, /workflows for running multi-agent work—get called automatically when you just talk to it in plain language. Only a few genuinely need you to type the slash, like /model to switch models, adjusting thinking depth, and /fast for fast mode. The whole atlas page is searchable and filterable, and downloadable straight from the blog.
What I’ve got for you today is the complete list of slash commands in Claude Code, as of the latest build.
I had it compile them into a web page: 91 commands total, each with a plain, beginner-friendly explanation you can actually follow. It’s hard to find a version this detailed anywhere, so share it, save it, work through it at your own pace. Leave a comment and I’ll get you the download link for the page.
Don’t panic: you don’t need to memorize all 91
A lot of people see 91 and immediately think, how am I supposed to remember all that? You genuinely don’t need to memorize them.
Why? Because as you use Claude Code, plenty of commands get called for you automatically the moment you say what you want in plain language. The context-related commands are like that; you never need to type them.
Take a recent one—the /goal command. What it does: you define an objective, the agent keeps working, and afterward it checks whether the objective was actually met, stopping only once it has been. In practice, though, plain language triggers it. You don’t have to type /goal every time.
Or the newest one, /workflows. Just say “spin up a parallel multi-agent task” and it calls that slash command on its own. So with this whole group, knowing it exists is enough—leave the rest to plain language.
The few actually worth typing out
That said, some commands you do want to fire off deliberately with a slash. The ones I use most:
- Switch models (
/model)—swap out the AI brain; - Switch how deeply the model thinks—same model, thinking shallower or deeper;
- Turn on fast mode (
/fast)—this one is genuinely satisfying; the whole loop of thinking and responding gets much faster, about 2.5x if I remember right.
One honest warning about fast mode, though: it does not run off your subscription plan. It draws straight from your Claude credit balance—real dollars. I remember it burning through $200 in about 20 minutes once, which was alarming. So go in knowing that.
One I use a lot: insights for a retro
One more I use often—the insights command. It runs a retro on how you’ve been using Claude Code lately:
- which projects you finished;
- which habits in how you talk to Claude Code are working well;
- and which problems you ran into, and how you might solve them—it really does offer suggestions.
It’ll even spot work you run frequently and suggest you pin it down as a skill, so it becomes more stable and more reliable.
Finally
The page is right there, go learn from it at your own pace. Short one today. See you next episode.
Source: EP0030_audio.mp3 · ASR model gemini-2.5-pro (chunked parallel) · full text of the roughly 3-minute original recording
[00:00] All right, friends, what I’ve got for you today is the complete guide to Claude Code’s slash commands, latest version as of today. There are 91 commands in here, with dead simple, beginner-friendly explanations and notes. You’d have a hard time finding a version this detailed anywhere online, so go ahead and share it, save it, work through it at your own pace. And of course, if you leave a comment, I’ll get you the download link for this page too.
[00:21] So in here, first you’ll see some of the commonly used commands. And don’t stress—you really don’t need to memorize all 91 slash commands. Why? Because some of these commands, in the course of using Claude Code, get triggered automatically when we just say what we want in plain language. So we can skip over those context commands.
[00:42] Take /goal, for example, which only showed up recently. It defines a goal, and after the agent finishes executing, it checks whether the goal was met, and only stops once it has been. With a goal like that, when we’re talking to Claude Code, plain natural language is enough to fire that command off—we don’t have to type out /goal every single time.
[01:07] Same with the ones below, and the rest you can read yourself. That goes for workflow too, the newest workflow command—as soon as we say “spin up a multi-agent parallel task,” it’ll call that slash command on its own.
[01:21] Of course there are also commands you do need to activate with a slash. The ones I’d say get used most are switching models, right, including switching the model’s thinking depth, and turning on fast mode. And fast mode really is a treat—the whole thing, the thinking and the getting back to us, goes really fast. I remember it’s something like 2.5x.
[01:47] But this one doesn’t run off our subscription plan—it charges straight to the balance in our Claude account, real hard dollars. I remember one time, in about 20 minutes, it burned through $200 of my credit. That’s terrifying.
[02:02] What other commands are there, let me think. Right, another one I use a lot is the insights command. That command can take how we’ve been using Claude Code over the recent stretch and do a retro on it: which projects we finished, what’s working, which habits in the way we talk to Claude Code are the good ones, then what problems we ran into and what approaches could solve those problems—it actually gives you suggestions on all of it.
[02:30] And for the stuff we run a lot, it’ll also suggest we turn it into a skill so it gets more stable and more reliable.
[02:40] So the page is right here, go work through it at your own pace. That’s it for today, very short, see you next episode, bye.