Instagram reels and short-form posts often promise a list of “secret ChatGPT commands” that supposedly unlock better answers. You may have seen prompts such as /handwritten, /humanize, /expert or /eli5. They look like software commands, but most are simply ordinary words preceded by a slash.
The distinction matters. ChatGPT does have some genuine interface shortcuts, but a viral prompt does not become an official feature merely because it starts with /. Here is how to tell the difference—and how to replace vague shortcuts with prompts that produce more dependable results.
Does /handwritten work in ChatGPT?
/handwritten is not a documented official ChatGPT command. If ChatGPT responds to it, the model is probably interpreting “handwritten” as a normal style instruction. The slash itself does not activate a hidden writing mode.
The result may therefore vary between chats, models and the text you provide. “Handwritten” is also ambiguous: it might mean casual prose, a handwriting-style image, or text that sounds less formal. A clearer instruction is more reliable:
Rewrite this in a natural first-person voice. Keep every factual claim unchanged, use varied sentence lengths, remove corporate jargon, and do not add personal experiences that are not in the original.
No prompt can guarantee that writing will be “undetectable” as AI-generated, and AI-detection scores are not a dependable measure of authorship. For publishing, the sensible goal is accurate, original and properly edited writing—not evading a detector.
Real commands versus viral prompt shortcuts
A genuine slash command is connected to the ChatGPT interface and normally appears in the suggestion menu when you type /. Availability can vary by platform, account and rollout, so the menu visible in your own composer is the safest current list.
OpenAI officially documents /canvas for opening Canvas, although its help page describes it as a backslash command in some places. OpenAI also documents /study where slash suggestions are available, while the more broadly supported shortcut is @study. Older macOS release notes mention slash shortcuts for actions including search, reasoning and image generation. These are interface actions—not secret instructions that improve every answer.
| Shortcut seen online | Official built-in command? | What usually happens | Better approach |
|---|---|---|---|
/handwritten |
No documented command | The model guesses the requested style | Describe the exact voice, audience and facts to preserve |
/humanize |
No documented command | May make prose more conversational, but results vary | Specify tone, sentence variety, jargon level and editing limits |
/eli5 |
Common internet shorthand, not a documented ChatGPT command | Usually interpreted as “explain simply” | Name the reader’s age or knowledge level and request an example |
/expert |
No documented command | Encourages an expert-like tone, not guaranteed expertise | State the domain, jurisdiction, assumptions and sources required |
/summarize |
Do not assume it is built in | Often treated as a normal request | Set length, format and what must not be omitted |
/canvas |
Officially documented | Opens or requests ChatGPT Canvas where supported | Use it for longer writing or coding work that benefits from editing |
/study |
Official where slash suggestions are available | Opens Study Mode after you select the suggestion | @study or the tools menu may be available more broadly |
A prompt formula that works better than a secret command
OpenAI’s own prompting guidance recommends being clear, specific and sufficiently detailed, then refining the request after reviewing the result. A useful reusable formula is:
Task + context + audience + constraints + output format + verification.
For example, “Compare these two phones” is workable, but the following is much more controlled:
Compare Phone A and Phone B for an Indian buyer with a ₹30,000 budget. Use current official specifications and India prices. Include a compact table, explain the display, performance, cameras, battery and update policy, flag any unverified claim, and finish with separate recommendations for gaming and photography.
Eight useful prompt templates to save
1. Natural rewrite without changing facts
“Rewrite the text for a general reader in a warm, direct tone. Preserve names, numbers, quotations and factual meaning. Vary sentence length, remove repetition and list every material change you made.”
2. Simple explanation
“Explain this to a reader who knows the basics but not the jargon. Use one everyday analogy, define technical terms on first use and finish with a three-point recap.”
3. Reliable summary
“Summarise this in five bullets of no more than 20 words each. Preserve dates, prices and warnings. Add a separate ‘What this does not say’ line.”
4. Product comparison
“Compare these products using only the supplied data and linked official pages. Create a table, name a winner for each category, separate claimed figures from measured results and explain who should buy each one.”
5. Fact-check a viral claim
“Check this claim against current primary sources. Separate confirmed facts, missing context and unsupported statements. Provide source links and publication dates. Say clearly when evidence is insufficient.”
6. Improve an email
“Rewrite this email to be concise and polite without sounding stiff. Keep the request and deadline explicit. Do not invent background information. Give me a subject line and the final email.”
7. Turn notes into an action plan
“Convert these notes into tasks. For each task include the owner, next action, deadline and dependency. Put unresolved decisions in a separate questions section.”
8. Research before writing
“Research this topic using current primary sources first. Show me the proposed angle and key facts with links before drafting. Avoid copying source wording and identify any claim that could not be independently verified.”
How to test a command before trusting it
- Type the slash and wait: if ChatGPT displays a selectable interface suggestion, it is likely a supported shortcut for your account.
- Check official documentation: social posts can remain online long after features change.
- Try it without the slash: if “humanize this” produces the same behaviour as
/humanize, you are seeing normal language interpretation rather than a special mode. - Inspect the result: a confident answer can still contain invented facts, citations or capabilities.
For recurring preferences such as tone, preferred units or response structure, ChatGPT’s Custom Instructions are more suitable than repeatedly pasting a made-up slash command. For high-stakes medical, legal or financial decisions, always verify the answer with qualified sources.
The bottom line
There is no universal catalogue of secret ChatGPT slash commands. A few slash shortcuts can open real tools such as Canvas or Study Mode on supported interfaces, but viral labels like /handwritten and /humanize are generally just compact prompts. They may appear to work because ChatGPT understands ordinary language—not because they unlock a hidden feature.
The smarter shortcut is a reusable prompt that clearly states the job, audience, constraints, format and verification standard. It takes a few more words, but it gives ChatGPT far less room to guess.
Sources: OpenAI Canvas guide, OpenAI Study Mode guide, OpenAI prompt-engineering best practices, OpenAI Custom Instructions guide and ChatGPT macOS release notes.
