ChatGPT Command Codes: 100 Hidden Shortcuts

100 slash-style triggers that switch ChatGPT into hidden modes — organized so you can actually find the one you need.

These aren't official features — they're behavioral triggers. Paste any of these into ChatGPT and it shifts how it thinks, writes, or responds. Some are prefixes, some are inline instructions, some restructure the whole output. Use them when the default response isn't what you need. Organized by use case so you can find the right one fast.

Output Format Control

/table — Present this as a comparison table with headers
/bullet — Rewrite this as a bulleted list, no prose
/numbered — Give me a numbered step-by-step, nothing skipped
/oneliner — Summarize in exactly one sentence
/tldr — Add a TL;DR block at the top before the full answer
/compress — Cut to half the length without losing key points
/expand — Double the depth on this — more examples, more detail
/split — Break this into two sections: What and Why
/grid — Format as a 2-column layout with labels on the left
/outline — Give me a structured outline, not prose
/headers — Reformat with H2 and H3 headers throughout
/json — Return this as valid JSON I can paste into code

Tone and Voice Switching

/formal — Rewrite this in professional, business-appropriate tone
/casual — Make this sound like I'm texting a smart friend
/direct — No hedging, no caveats, just the answer
/blunt — Tell me the hard truth, skip the softening
/warm — Add empathy without losing the substance
/academic — Use precise language, cite reasoning, structured arguments
/punchy — Short sentences. High contrast. No filler.
/authoritative — Write like a recognized expert in the field
/conversational — First person, loose structure, sounds like talking
/neutral — Strip all opinion, state only facts and options
/provocative — Take a strong stance, make it interesting
/socratic — Ask me questions instead of giving answers

Role and Persona Modes

/expert — Respond as a recognized expert in [field], not a generalist
/critic — Play devil's advocate, find every flaw in this
/advisor — Act as my trusted advisor, give me your honest recommendation
/editor — Review this and mark what's weak, unclear, or redundant
/interviewer — You're interviewing me about this topic, ask sharp questions
/coach — Guide me through this step by step, ask before moving on
/copywriter — Rewrite this to convert, not just inform
/engineer — Think through the technical constraints and tradeoffs
/strategist — Zoom out and tell me the play, not the tactic
/investor — Evaluate this like you're deciding whether to fund it
/operator — How would someone actually implement this, practically

Thinking and Reasoning Modes

/stepbystep — Walk through your reasoning before giving the answer
/firstprinciples — Break this down to its most basic assumptions
/pros-cons — Give me a balanced list before recommending anything
/assumptions — List every assumption baked into this before continuing
/steelman — Make the strongest possible case for the opposite view
/debug — Find the logical errors or gaps in my reasoning
/contrarian — What would a smart skeptic say about this?
/secondorder — What are the downstream consequences, not just the obvious ones?
/tradeoffs — What am I giving up with each option?
/falsify — What would prove this wrong?
/bottleneck — What's the single constraint limiting this the most?

Writing and Editing Commands

/rewrite — Rewrite this from scratch, same meaning, better execution
/tighten — Cut every word that isn't earning its place
/vary — Give me 3 different versions of this
/headline — Write 10 headline options for this content
/hook — Write 5 different opening lines that pull the reader in
/cta — Write 5 calls to action for this offer or piece of content
/transitions — Smooth out the transitions between these sections
/passive-fix — Find every passive sentence and rewrite it as active
/jargon — Flag every piece of jargon and suggest plain-language swaps
/concise — Rewrite this at a 6th-grade reading level without dumbing it down
/readability — Score this and tell me exactly what's making it hard to read

Research and Analysis Commands

/summarize — Summarize the core argument in 3 sentences
/extract — Pull out only the key facts, ignore everything else
/compare — Side-by-side comparison of [A] vs [B]
/gaps — What's missing from this analysis?
/sources — What types of sources would back up these claims?
/patterns — What patterns do you see across these examples?
/simplify — Explain this like I'm smart but have no background in it
/define — Define the core terms in this before explaining it
/context — Give me the background I need to understand this properly
/timeline — Lay this out as a chronological sequence of events
/map — Create a mental model or framework for understanding this

Business and Marketing Commands

/icp — Define the ideal customer profile for this offer
/positioning — Write a one-line positioning statement for this
/objections — List the top 5 objections a buyer would have and how to handle them
/funnel — Map out a simple funnel for this offer from cold to close
/email-sequence — Draft a 5-email nurture sequence for this topic
/angles — Give me 7 different angles I could write about this topic
/offer — Build an irresistible offer structure around this product
/pain — What are the real underlying pains my buyer has before they find me?
/value-prop — Rewrite my value proposition to be specific and measurable
/story — Turn this feature or result into a short customer story
/script — Write a 60-second talk track for this situation
/case — Frame this as a before/after case study structure

Productivity and Workflow Commands

/action — Convert this into a concrete next-action list
/priority — Rank these by impact, not urgency
/delegate — Tell me what in this I should hand off and to whom
/automate — What parts of this workflow could be automated and how?
/sop — Turn this into a standard operating procedure with clear steps
/checklist — Convert this into a pre-flight checklist format
/template — Turn this into a reusable template I can fill in
/batch — How do I do this 10x at once instead of one at a time?
/timeline — Build a realistic timeline for completing this
/bottleneck — Where will this break down and what do I do about it?
/decision — Help me make this decision using a clear framework

Code and Technical Commands

/explain-code — Walk me through what this code does line by line
/refactor — Clean this code without changing what it does
/debug — Find the bug in this and explain why it's happening
/comment — Add clear inline comments to this code
/test — Write unit tests for this function
/optimize — Make this faster or more efficient, and tell me what you changed
/docs — Write the documentation for this function or API endpoint
/convert — Convert this from [language A] to [language B]
/regex — Write a regex pattern that matches [description]
/schema — Design a database schema for this use case
/api — Draft the API spec for this feature in plain language

Conversation and Session Commands

/context — Here's the background you need before we continue: [paste it]
/constraint — Only give me answers that work within [these limits]
/audience — Everything from here forward is written for [describe audience]
/goal — My actual goal is [X] — keep that in mind as we go
/format-lock — Use this exact format for every response in this session
/no-caveats — Skip all disclaimers and qualifications unless safety-critical
/iterate — Take your last response and make it 20% better
/different — Give me a completely different approach, not a variation
/pushback — Tell me what's wrong with my idea before helping me build it
/done — Stop explaining and give me the final deliverable only
/restart — Ignore the last few messages and start this fresh with new direction

Stack commands to combine effects. '/blunt /bullet /expert' at the top of a prompt gives you a ruthless, scannable answer from a domain expert. Test which combinations your workflow actually needs and build a shortlist you use daily.

These only work as well as the prompt they're attached to. A command changes how ChatGPT responds — it doesn't fix a vague question. Know what you want, use the command to shape the delivery, and you'll cut the back-and-forth in half.