← The AI Hype Audit — all 181 verdicts

PARTLY “35 CHATGPT CODES — powerful prompts, better answers” — they work; they're just not codes Facebook · Aug 17, 2026

The claim“35 CHATGPT CODES — Powerful prompts. Better answers.” — ELI5, TL;DR, SWOT, Act as [Role], Feynman, Socratic, Devil's Advocate, compare-in-a-table, JSON/Markdown output, and thirty more.

Every item on the card is a legitimate, working prompting pattern: ELI5, TL;DR, step-by-step, compare-in-a-table, pros/cons, SWOT, role prompting, the Feynman technique, Socratic questioning, devil's advocate, and output-format requests all genuinely steer any LLM — ChatGPT, Claude, or Gemini — toward more useful answers. As a free cheat sheet for beginners, it's fine. The problem is the word “codes.” There is no hidden command layer — these are plain-English phrases the model understands the same way it understands any sentence. Nothing here is documented as a command or parameter anywhere in OpenAI's or Anthropic's docs, nothing is ChatGPT-specific, and nothing is secret. The framing trains beginners to think prompting is incantation instead of clear instruction — which is precisely the mental model that keeps people bad at prompting.

What holds up

  • All 35 patterns genuinely work — formats, roles, and reasoning frames measurably improve output on any major LLM
  • It's free with no comment-gate — just a shareable cheat sheet
  • Nothing on the card is wrong, only mislabeled

What doesn't

  • They are not “codes” — no OpenAI or Anthropic doc lists any of these as commands or parameters; they're ordinary sentences
  • “Secret unlock” framing teaches incantation-thinking, the exact opposite of the real skill
  • Nothing is exclusive, new, or ChatGPT-specific — the same phrases work identically in Claude or Gemini

The catch

Calling plain English “codes” manufactures scarcity around something that was never scarce — the model understands “explain this simply” because it understands English, not because you found a cheat key.

How to actually do it

  • Compress all 35 into the one real skill: tell the model the audience, the format, and the reasoning style you want — “Explain X to a 10-year-old, then give me a pros/cons table”
  • Keep three or four favorites in muscle memory (ELI5, table comparison, devil's advocate, output format) — that covers most daily use
  • When a card promises secret codes, magic words, or hidden commands for an LLM, downgrade the source — it signals the author doesn't know how these models work

A perfectly good beginner cheat sheet mislabeled as a cheat code — use the patterns, unlearn the framing.

Confidence
High
Posted by
an AI-prompts page recycling prompting cheat sheets as infographics

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.

© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.