What most "Claude cheat sheets" get wrong
Search "Claude prompt cheat sheet PDF" and you will find dozens of them. Most are:
- Lists of 50 to 100 prompt codes copied from Reddit threads
- Authored by someone who never tested whether the codes actually work
- Heavy on ULTRATHINK, GODMODE, ALPHA (codes that mostly test as placebo)
- Missing when-NOT-to-use warnings and failure modes
- Stale, written once and never updated as Claude models change
I spent three months running 120 prompt codes through a controlled test to figure out which ones do anything. The short version is that about 5 codes actually shift the reasoning, roughly 40 change the output shape in useful ways, and the rest are formatting drift dressed up as "secret modes." That is the frame I brought to this comparison.
Useful cheat sheets, free or paid, share three features:
- Each code includes before and after proof (actual output text showing the difference)
- Categorized by what they do (reasoning vs format vs tone)
- Include when-NOT-to-use warnings (because most codes backfire in some contexts)
Here is a breakdown of what is actually worth downloading.
Free options that are real
The CLSkills free prompts library
We maintain a browsable free library of 100 Claude prompt codes at clskillshub.com/prompts. Organized by category (writing, reasoning, coding, research, communication) with one-line descriptions and copy-to-clipboard buttons. No email required, no PDF gate.
Good if you want to browse and experiment. Not tested-and-classified like the paid version.
The 75-page Claude guide
Full beginner-to-advanced Claude guide covering setup, the CRISPE framework for structured prompting, MCP servers, agents, and 20 must-know prompt codes with examples. Free PDF, no signup barrier beyond an email address.
Better for "I am new to Claude and want the basics" than for "I want the tested codes."
The paid option
CLSkills Cheat Sheet
Three tiers, and I will tell you honestly what is in each:
- Lite ($10): 50 codes with before and after examples, markdown download. Good if you want the tested set without paying for the analysis.
- Full ($15): All 160+ codes plus when-NOT-to-use warnings, code combos, and 10 workflow playbooks. This is the tier most buyers pick because it is the practical daily driver.
- Pro ($35): Everything in Full plus the Deep Dive analysis on the highest-value codes (the actual test data, classification, and failure modes for the roughly 5 codes that showed a real reasoning shift). Also includes 20 bonus reusable prompt templates and a 30-day direct email channel with me.
What makes it different from the Gumroad and Notion versions: every code has been tested against a baseline (same prompt, no prefix) and classified. The before and after examples in the Pro tier use actual Claude output on the exact test prompts I ran, not simulated examples.
Digital product, one-time payment, no subscription, lifetime updates. When a new Claude model ships, the tests get rerun and your unlock URL serves the updated version.
Other paid options
There are a few decent paid cheat sheets on Gumroad and Notion templates in the $5 to $20 range. The quality bar to check for is simple: do they publish before and after outputs? If the cheat sheet is just a list of codes with descriptions and no actual Claude outputs showing the effect, it is probably not tested.
Questions to ask before paying anyone (including us):
- Do you have test data showing these codes work?
- Can I see an example of a before and after for any specific code?
- When was this last updated? Which Claude model was it tested on?
- Do you include warnings about when NOT to use each code?
If the author cannot answer these, you are buying a list of Reddit copy-paste, not a tested reference.
Free library vs paid Cheat Sheet: which do you need?
Free library (clskillshub.com/prompts) is enough if:
- You want to experiment with 100 codes and figure out which ones work for YOU
- You have time to test each code yourself
- You do not need failure mode warnings or combo strategies
- You are new to Claude prompting and want to explore
Paid Cheat Sheet is worth it if:
- You want the reasoning-shifter classification (which 5 codes actually matter)
- You want to skip the placebos without testing each one yourself
- You want the paste-ready combo templates for common tasks (cold email, debug, summarize)
- You want the lookalike-placebo warnings (codes that look similar but do not work)
- You value 30-day direct email access to me (Pro tier)
FAQ
Do I need a PDF or is the web version enough?
Both are included with the Cheat Sheet. The markdown file is attached to your purchase email and works offline in Obsidian, Notion, VS Code, or any markdown reader. The live web version updates when new codes are tested. Your unlock URL always serves the latest.
What is the difference between the guide and the cheat sheet?
The guide teaches Claude from scratch (setup, frameworks, MCP, agents, how to write prompts). The cheat sheet is the deep reference for prompt codes specifically. Guide equals the education. Cheat sheet equals the lookup table. They are complementary, not redundant.
Does this work with Claude Haiku 4.5?
Yes. The cheat sheet was re-tested when Haiku 4.5 shipped. About 20% of rankings shifted. The headline codes (L99, /skeptic, /ghost) held up cleanly.
Is there a free Claude cheat sheet PDF?
Yes. The 100-code prompts library at clskillshub.com/prompts is free and browsable. If you want a single downloadable PDF of the full 160+ tested codes with classification and test data, that is the paid Cheat Sheet starting at $10.
What is your refund policy?
Digital product, one-time payment, all sales are final. If something genuinely did not land for you, email team@clskills.in and I will add you to the full Skills Library (lifetime access) as a goodwill gesture. If something is technically broken (dead link, wrong tier, missing access), I fix it directly.
Can I use this with ChatGPT or Gemini?
Most of the codes are Claude-specific. Roughly 30% of the format-only codes (/raw, /bullets, /json) transfer cleanly. The reasoning-shifters (L99, /skeptic) are Claude-specific because they rely on how Claude handles meta-instructions. A tested multi-model version is on the roadmap.
The real filter
If someone is selling you a Claude cheat sheet and cannot show you test data, skip it. The useful cheat sheets are the ones where the author actually ran the codes against baselines and published the methodology.
Where the rest of this lives
If you liked this comparison and want the full tested set, the Cheat Sheet is where the rest of this lives. 160+ codes with the exact temperature and top_p I use for each, before-and-after outputs on the ones that showed signal, and 10 workflow playbooks. $15 for Full, $35 for Pro with the verified-test analysis.
The free 75-page Claude guide covers the basics if you want the long version first: clskillshub.com/guide