I tested popular GPT shortcuts: here's what works on real tasks

When you work with GPT every day, the first thing you want is to speed things up. That's why people came up with shortcuts: short commands like EL5 or /BULLET that tell the model what style and priorities to use in its answer. Over a month, I tested the most popular ones on real tasks: reviewing code, simplifying articles, and finding flaws in logic. The results were interesting: not all shortcuts are equally useful.

The workhorses

EL5 (Explain Like 5) turns any complex concept into an explanation a five-year-old could follow. In practice, that means no jargon, just the core idea and concrete examples. It works reliably in 95% of attempts, which is a high rate for a prompt tool. It's a lifesaver when you need to get up to speed on something unfamiliar or explain it to a colleague without extra words.

/BULLET formats the answer as a bulleted list. It never fails and runs like clockwork. It's simple, and the structure is visible at a glance.

/CHAIN lays out step-by-step reasoning. It's slower than the others, but it's the most reliable option for logic problems.

• EL5 — simplifies complex ideas down to the basics
• /BULLET — formats output as a bulleted list
• /CHAIN — step-by-step reasoning for logic
• /SCIENCE — backs claims with facts and research

Shortcuts that work only half the time

/REDTEAM asks the model to critique an idea and look for weak spots. It sounds useful, but in practice the output varies: anything from constructive criticism to petty nitpicking, depending on the day and the model version.

/LAWYER analyzes things like a lawyer would: it looks at context, hunts for contradictions, and flags risks. It does fine with standard situations, but on complex questions it can miss the main point and get stuck on details. It's useful as a first-pass check, but not as a final answer.

What to avoid entirely

/JOKE, a command to add a joke, is so unreliable that you should forget it exists. I tried it ten times in a row: five times it was weird and out of place, three times it wasn't funny, and twice it was fine. That's 20% usefulness for 100% of the effort.

/METAPHOR and /HAIKU look nice on paper, but in real work they cost more time than they save.

What this means in practice

It's better to master three or four reliable shortcuts than to chase the perfect one for every situation. The main discovery: shortcuts combine well. /BULLET + EL5 together work better than either one alone. And don't trust promises about tools: test them yourself on your own tasks for two weeks, then decide.