Claude Prompt Shortcuts Cheat Sheet
Searchable, copy-ready prompt shortcuts for Claude — fact-checked so you know which are just phrasing conventions and which are real Anthropic-documented techniques.
/ACT, /TLDR-style
commands. These are prompt-prefix conventions that work because Claude follows plain-language
instructions — typing the slash isn't required, and Claude doesn't parse it as special syntax.
Claude Code (the CLI) has its own, unrelated set of real slash commands — see the
comparison table below.
No shortcuts match your search.
Real Claude Code slash commands
These are genuine built-in commands in the Claude Code CLI — a different tool from claude.ai's
web chat, and a different command set from the prompt shortcuts above. The full list varies by
version; type / inside Claude Code to see yours.
| Command | What it does |
|---|
How it works
"Shortcut" is a loose word here — none of these change how Claude parses your message. They're prompt-prefix conventions: short, memorable phrasings people use to reliably get a certain kind of answer (a summary, a role, a specific format) without re-writing the same instruction from scratch every time. The value is in having a consistent phrasing you can paste, not in a hidden feature you're unlocking.
Two groups on this page are worth telling apart. Most of the 29 shortcuts are community conventions — practical, but not documented anywhere by Anthropic. A separate "Anthropic-verified extras" group (XML tags, multishot examples, extended thinking) comes straight from Anthropic's own prompt-engineering documentation, and those are worth reaching for first when a community convention and a documented technique overlap — extended thinking, for instance, is a real setting that changes how much Claude reasons, while "deliberate thinking" is just asking nicely in text.
Click Copy on any card to grab a ready-to-paste prefix — drop your actual content after it.
FAQ
Does Claude actually have slash commands like /ACT or /TLDR?
Not in claude.ai's web chat. Typing "/TLDR" works only because Claude reads it as a plain-language instruction to summarize — the same result comes from writing "summarize this" with no slash at all. Claude Code, Anthropic's separate CLI tool, does have real built-in slash commands, but they're a different set entirely (/clear, /compact, /model, and so on) and control the coding session, not the model's writing style.
So do these shortcuts actually work?
Yes — as instructions, not as commands. Claude follows natural-language directions well, so a prefix like "Act as a senior engineer" or "Explain this like I'm five" reliably shapes the response. The slash is just a memorable label; nothing breaks if you leave it off or write the instruction in full sentences.
What's the difference between these and Claude Code's slash commands?
These shortcuts are phrasing conventions you type inside a message to steer Claude's tone, structure, or reasoning. Claude Code's slash commands are real CLI features — typing /clear or /compact does something the chat interface itself executes, unrelated to how Claude writes its response. See the comparison table on this page.
Which of these are official Anthropic techniques?
XML tags, multishot examples, and extended thinking are documented in Anthropic's own prompt-engineering guide and are marked "Anthropic-verified extras" on this page. The rest are community conventions — genuinely useful, but not sourced from Anthropic documentation.
Is prefilling Claude's response still a recommended technique?
Not as a general-purpose trick going forward. Prefilled assistant messages are no longer supported on newer Claude models (Claude 4.6 and later reject them), and Anthropic notes that improved instruction-following has made most prior use cases for prefill unnecessary. It's intentionally left off this list.