Claude Code Slash Commands: 10 Hidden Gems for Beginners
The 10 Claude Code slash commands I actually use every day — /rewind, /compact, /btw and friends — explained for non-programmers, with zero jargon.
Evgenii Arsentev · PhDReviewed for accuracy by Evgenii Arsentev, PhD · 2026-10-07
Claude Code slash commands are short instructions that start with / and control the session itself: switch models, undo changes, clean up memory, check your bill. You don't need to memorize them — type / and a menu appears — but knowing ten of them by heart turns you from a tourist into a local.
I'm Evgeny, and I teach people who have never written a line of code to build things with Claude Code. Every tutorial shows you /init and /help and stops there. Meanwhile the commands that actually save your evenings — the undo button, the side-question trick, the context vacuum cleaner — sit quietly in the menu, unmentioned. Let's fix that.
Where do I find the full slash commands list?
Type a single / in the prompt and Claude Code shows everything available to you: built-in commands, skills, and anything added by plugins. Keep typing letters to filter — /re narrows it to /resume, /rewind, /recap and so on. One detail beginners miss: a command only counts at the very start of your message. Anything you type after the command name is passed to it as arguments.
The 10 commands worth learning by heart
My daily ten, in the order you'll need them
- 1/help — shows help and the available commands. The honest starting point.
- 2/clear — starts a fresh conversation with empty context. The old one stays available in /resume.
- 3/resume — reopens a past conversation by name or from a picker. Alias: /continue.
- 4/rewind — rolls the conversation and/or your files back to an earlier point. Aliases: /undo, /checkpoint.
- 5/compact — summarizes the conversation to free up context when Claude starts forgetting things.
- 6/model — switches the AI model and saves it as your default for new sessions.
- 7/usage — shows session cost, plan limits, and activity stats. Aliases: /cost, /stats.
- 8/btw — asks a quick side question without adding it to the conversation history.
- 9/export — saves the whole conversation as plain text, to clipboard or a file.
- 10/doctor — checks your installation and settings; it lists what it found first and asks before fixing anything.
- 11/cd — moves the session to a different working directory. The context survives the move: your new folder's CLAUDE.md is appended as a message instead of restarting everything. Handy when you start in one project and realize you need to work in a related one.
Two commands worth a closer look: /rewind and /btw
/rewind
Claude rewrote three files and broke your page? /rewind opens a menu where you restore the code, the conversation, or both to a previous point. Pressing Esc twice on an empty prompt opens the same menu. This command alone is why I let beginners experiment fearlessly.
/btw what was the name of that config file again?
/btw answers from what's already in the conversation, shows the reply in a small overlay, and never enters the history. Perfect for "wait, what did we just do?" moments mid-task.
/cd ../other-project
/cd switches the current session to a different working directory without wiping your conversation. The prompt cache is preserved: the new directory's CLAUDE.md is appended as a message, not a system-prompt rebuild. Claude will ask if you trust the new directory first, so there are no silent access surprises. If you want to give Claude access to another folder without actually moving there, use /add-dir instead.
Do slash commands take arguments?
Many do. Whatever you type after the command name becomes its input: /compact focus on the database changes tells the summary what to keep, /clear my-experiment names the conversation you're leaving so you can find it later in /resume, and /plan fix the login page jumps straight into plan mode with a task. No flags, no syntax to study — plain words work.
Slash commands vs skills: what's the difference?
Some entries in the / menu are built-in commands (they change settings or session state), and some are skills — prepared prompts that get handed to Claude, which Claude can also trigger on its own when relevant. /loop and /code-review are skills, /model and /clear are commands. For everyday use the difference barely matters: you type /, pick a thing, it works. Run /skills to see which skills you have.
How to create custom slash commands
Yes — your own entries in the / menu are made as skills: reusable prompts you write once and invoke by name. If you find yourself typing the same request every morning ("check the site, summarize errors, suggest fixes"), that's a skill waiting to be born. Ask Claude itself to create one for you; it knows where the files go.
Beyond slash: three tricks without a /
Start a message with ! to run a terminal command directly, without asking Claude. Type @ to autocomplete a file path into your prompt. Press Ctrl+R to search everything you've ever typed. None of these are slash commands — which is exactly why nobody mentions them in tutorials.
Lab: a five-minute command tour
0/5Here's the kicker: run /powerup once this week. It's a built-in set of quick interactive lessons with animated demos — Claude Code teaching you Claude Code. Ten minutes there, plus the ten commands above, and you'll navigate sessions better than most people who've used the tool for months.
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →References
Selected peer-reviewed papers and preprints on the AI research behind this topic.
- [1]Wang et al. (2024). Learning From Failure: Integrating Negative Examples when Fine-tuning Large Language Models as Agents. arXiv:2402.11651
- [2]Rothfarb et al. (2025). Hierarchical Multi-agent Large Language Model Reasoning for Autonomous Functional Materials Discovery. arXiv:2512.13930
- [3]Estrada et al. (2021). AIRCC-Clim: a user-friendly tool for generating regional probabilistic climate change scenarios and risk measures. arXiv:2111.01762
- [4]Ding et al. (2025). ToolRegistry: A Protocol-Agnostic Tool Management Library for Function-Calling LLMs. arXiv:2507.10593
- [5]Xu et al. (2023). WizardLM: Empowering large pre-trained language models to follow complex instructions. arXiv:2304.12244
- [6]Yin et al. (2024). MuMath-Code: Combining Tool-Use Large Language Models with Multi-perspective Data Augmentation for Mathematical Reasoning. arXiv:2405.07551
- [7]Wang et al. (2023). Demystifying Instruction Mixing for Fine-tuning Large Language Models. arXiv:2312.10793
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Reading is a good start
Want to actually build this?
Guides explain. The free course transforms — personalized, gamified, and built to get you shipping fast.
Start the free course →