Claude Cheat Sheet

You are working with a Claude reference covering Projects, artifacts, file handling, search, extended thinking and skills, plus 25 one-line prompts for writing, analysis, code and learning. You have been activated to use the other nine-tenths of what Claude does.

CLAUDE CHEAT SHEET Most people use about a tenth of what Claude can actually do, then blame the model when the answers come back average. The gap usually isn't the model - it's that Projects, artifacts, extended thinking, and skills each change what you get back, and almost nobody reads far enough to find out how. This is the reference for the other nine-tenths: what to turn on, what to say, and when each feature actually matters. WHAT EACH FEATURE ACTUALLY DOES Projects - A Project is a persistent workspace with its own instructions and files, shared across every conversation inside it. Use one per ongoing piece of work (a client, a codebase, a course) instead of re-pasting the same background every time you open a new chat. The instructions field is read on every message in that Project - put your standing rules there (tone, format, constraints) instead of repeating them. Artifacts - A side panel that holds a standalone piece of output - code, a document, a diagram, an interactive page - separate from the conversation. Ask for one explicitly ("as an artifact") when the output is something you'll want to view, edit, or reuse on its own rather than read once in the chat. Claude can update an artifact in place across turns, so iterate by asking for the specific change rather than requesting the whole thing again. File handling - Claude reads PDFs, images, spreadsheets, and text files you upload, and can reason over their actual content, not just a filename. For long or many documents, be specific about what to extract - "list every date, party, and obligation" gets a usable table; "summarize this" gets a paragraph you'll have to re-read the source to verify. Search - Claude can look things up on the web when the answer depends on something current or something outside its training. It's most useful when you ask it to verify a specific claim or find a specific current fact, not as a general substitute for asking a clear question. Extended thinking - For harder reasoning, analysis, or multi-step problems, Claude can think through the problem before answering. It helps most on genuinely hard tasks (nontrivial math, multi-constraint planning, subtle bugs) and adds little on simple ones - save it for when the task actually needs the extra step-through. Skills - A skill is a reusable, packaged set of instructions Claude can load automatically when a task matches it, instead of you pasting the same detailed prompt every time. Worth building once you're repeating the same multi-step request more than a few times. 25 ONE-LINE PROMPTS Writing 1. "Rewrite this in [X] words, keeping every number and name exact." 2. "Give me three headline options ranked by how much they promise versus how much they're willing to risk sounding plain." 3. "Cut this by a third without losing any concrete detail - only cut abstraction and hedging."