Learn / Commands
Claude Code command cheat sheet
A command is a shortcut that tells Claude exactly how to respond, so you stop re-typing the same instructions. A few are built into Claude Code. Most you set up once, by saving a short instruction as a command, then reuse forever. Here are the ones worth knowing, grouped by what you are trying to get done.
Two kinds of commands
Some are built into Claude Code and work the moment you open it. The rest you set up once. In Claude Code, save the instruction as a file in your ~/.claude/commands folder and name the file after the command, so brutal.md becomes /brutal. A custom command and a skill are now the same idea, so saving it as a skill works too. Prefer Claude.ai? Paste the same instruction into Settings → Personal Preferences instead. Either way: copy the instruction, set it up once, and the command is yours.
Built into Claude Code
These work the moment you open Claude Code. Nothing to set up, just type them.
Tells Claude to spend its full thinking budget before answering. Worth it for strategy, tricky decisions, or anything where the first answer usually misses something. It is a word you drop into your message, not a slash command.
Just type it in Claude Code. No setup needed.
Switches Claude into plan mode before a big change. It proposes a step-by-step plan and waits for your go-ahead before touching anything. Type it at the start of your message.
Just type it in Claude Code. No setup needed.
Summarises a long conversation so Claude stays sharp instead of losing the thread. You can steer what it keeps, for example asking it to hold on to every decision about your brand voice.
Just type it in Claude Code. No setup needed.
Wipes the conversation and starts fresh. Use it the moment you switch tasks, so the last thing you worked on stops bleeding into the next one.
Just type it in Claude Code. No setup needed.
Switches the underlying model mid-session. Reach for the most capable model when the thinking is hard, and a faster one when you just want quick drafts.
Just type it in Claude Code. No setup needed.
Reopens an earlier session so you can carry on exactly where you left off, instead of re-explaining everything from the start.
Just type it in Claude Code. No setup needed.
Sharpen your thinking
Most people use AI to agree with them. These do the opposite.
Honest advisor mode. Claude drops the "great question" padding and tells you what is actually weak about your plan.
When I type /brutal, drop all flattery and soft framing. Tell me the strongest objection to what I just said, the one assumption I am leaning on most, and the thing I am avoiding. Stay kind, but do not soften the diagnosis. Attacks your plan the way a sceptical investor or a competitor would, so you find the holes before they do.
When I type /redteam followed by a plan or idea, switch to adversarial reviewer. List the three weakest assumptions in order of risk, the failure a hostile critic would attack first, and the one question that changes everything if the answer is wrong. Imagines the plan already failed twelve months from now, then works backwards to explain exactly why.
When I type /premortem followed by a plan, assume it is a year from now and the plan has failed. Tell me the most likely cause, the warning sign I ignored at the start, and the first domino that fell. Give me a specific story, not a generic list of risks. Strips an idea back to what is actually true and rebuilds from there, instead of inheriting assumptions you never chose.
When I type /firstprinciples followed by a topic, separate what is genuinely true from what is just convention I have inherited. Throw out the conventions, rebuild the answer from the truths, and end with the one question that would change the whole conclusion. Write and edit faster
Turn a rough draft into something you would actually publish.
Rewrites text so it reads like a person wrote it. No em dashes, no preamble, varied sentence length, none of the usual AI tells.
When I type /ghost followed by text, rewrite it to remove every AI tell. No em dashes, no "I hope this helps", no balanced on-one-hand pairs, no "in conclusion". Vary the sentence length, use fragments where they read naturally, and give me the rewrite only. Cuts a draft by a third without losing a single point. Kills the filler, keeps the substance.
When I type /tighten followed by text, cut it by at least 30 percent without losing any claim, number or example. Delete filler, merge repetitive lines, and move to active voice. Show me the word count before and after. Adds energy and specificity to flat copy without changing what it means.
When I type /punchup followed by text, sharpen it. Swap weak verbs for strong ones, cut filler words, replace vague adjectives with specifics, and tighten it by about 20 percent. Keep my meaning and my voice intact. Turns a one-line bullet into a full paragraph with an example and the obvious objection answered.
When I type /expand followed by a bullet or rough idea, develop it into one paragraph. Open with the core point, add a concrete example, answer the obvious objection in a sentence, and match my voice. Around 100 to 150 words. Make content
Get from a topic to something postable, fast.
Generates ten opening lines for a post across proven angles, so you stop staring at a blank first line.
When I type /hook followed by my topic, write ten opening lines across five angles, two each: contrarian, specific number, painful truth, before and after, and curiosity gap. Label each angle and keep every hook under fifteen words. Takes one piece of content and reshapes it into a tweet, a LinkedIn post, an email subject line, a carousel hook and a video hook.
When I type /repurpose followed by content, reshape it into five formats: a tweet under 270 characters, a LinkedIn post opener, an email subject line under 50 characters, an Instagram carousel cover line, and a short-form video hook. Label each one. Turns a topic into a swipeable carousel outline: a cover line, one idea per slide, and a closing prompt.
When I type /carousel followed by a topic, build a seven-slide carousel outline. Slide one is a bold cover line under eight words. Slides two to six are one idea each, with a short title and a couple of supporting lines. Slide seven is the takeaway plus a follow prompt. Writes seven different calls to action for the same offer, from low-friction to direct, so you can pick the one that fits.
When I type /cta followed by my offer, write seven calls to action across these flavours: low-friction, curiosity, urgency, social proof, identity, direct, and risk-removal. Keep each one under twenty words. Learn and decide
Understand something properly, or make a call with less second-guessing.
Explains a concept like you are smart but new to it. One everyday analogy, no jargon.
When I type /eli5 followed by a concept, explain it as if to a curious beginner. Use one everyday analogy as the anchor, no jargon, short sentences, and end with the next question worth asking. Keep it under 100 words. Compresses anything long into a one-sentence thesis, three key points, and what to do about it.
When I type /tldr followed by content, give me a one-sentence thesis, three bullet points of the most important ideas, and one sentence on what I should do or think differently as a result. Under 100 words, no preamble. Lays two or more options side by side on the dimensions that actually matter for your decision, not generic ones.
When I type /compare followed by two or more options, build a side-by-side on the five or six dimensions that matter for this specific decision. Score each option, add a best-for line for each, and end with one recommendation and the condition that would change it. Weighs a decision with each point ranked by how much it actually matters, instead of a flat list.
When I type /pros-cons followed by a decision, give me a weighted analysis. For each point, note why it matters in my context and rate its impact high, medium or low. End with the one pro that outweighs everything and the one con that kills the deal if it turns out to be true. Plan and ship
Turn a fuzzy idea into something you can hand off or act on.
Turns a messy idea into a tight one-page brief that a collaborator, a client, or future-you will actually read.
When I type /brief followed by a topic, write a one-page brief: the objective in one sentence, two or three lines of context, the key decisions to make, the constraints, and the single next step. Keep it under 300 words. Converts a plan or process into a step-by-step checklist you can actually run.
When I type /checklist followed by a goal or process, turn it into an ordered checklist. Each item starts with a verb and is one specific action. Flag anything that blocks other steps. Keep each line short enough to copy and paste. Turns a topic into a structured outline ready to write from: sections, subsections and key points.
When I type /outline followed by a topic, return a structured outline. Main sections, two or three subsections each, a few bullet points per section noting what to cover, plus an intro note and a conclusion note. Keep the section titles punchy. Writes a complete email, subject line to sign-off, calibrated to who it is going to.
When I type /email followed by the goal and recipient, write the whole email: a subject line under 50 characters, a personalised opening, a body of two or three short paragraphs, one clear call to action, and a sign-off. Match the formality to the relationship. Commands are the quick wins. The real leverage is building your own tools. I run hands-on workshops where non-technical people build and deploy a real website with Claude Code in one day.