01

Claude Basics

Select a model to see best use cases
H
Claude Haiku
Fast & cheap
S
Claude Sonnet
Balanced performance
O
Claude Opus
Most capable
Click a model above
Use cases will appear here

"Claude's model family isn't about power alone — it's about picking the right tool for speed, cost, and capability."

Anthropic's Claude comes in three tiers. Haiku is the fastest and cheapest — great for high-volume, simple tasks like classification or summarization. Sonnet is the daily driver — balanced performance for writing, analysis, and coding. Opus is the most capable — for complex reasoning, research, and tasks where quality matters more than cost. Newer versions (Claude 3.5, Claude 4) bring improvements across all tiers.

Claude model family: Haiku (fast/cheap) → Sonnet (balanced) → Opus (powerful). Choose based on task complexity, latency requirements, and cost constraints.
02

Skills and Plugins

Toggle skills on to give Claude new abilities
🔍 Web Search
⚡ Code Runner
📄 PDF Reader
🧮 Calculator
🎨 Image Gen
📅 Calendar
Claude now has access to: —

"By itself, Claude is a reasoning engine. Skills are what connect that reasoning to the real world."

Claude Skills (also called plugins or tools) extend Claude's capabilities beyond text generation. Web Search lets Claude access current information. Code execution lets it run Python, check its math, and test solutions. File upload lets it read PDFs and spreadsheets. Calendar and email integrations let it take actions. Each skill gives Claude a new "hand" to interact with the world.

Claude Skills: Modular capabilities that extend Claude beyond text generation — including web search, code execution, file reading, and third-party app integrations.
03

MCP — Model Context Protocol

Click a tool to see the connection light up
Claude
AI Model
MCP Server
Protocol Layer
📁
File System
🌐
Web Browser
📝
Notion
Select a tool above
Click any tool node to see how Claude connects to it via MCP.

"MCP is like USB-C for AI — a single standard that lets any AI connect to any tool."

The Model Context Protocol (MCP) is an open standard created by Anthropic that defines how AI models connect to external tools and data sources. Instead of building custom integrations for each tool, developers build one MCP server that any compatible AI can use. This means Claude can talk to your file system, browser, Notion workspace, or custom databases — all through the same protocol. MCP is becoming the standard for AI-tool connectivity.

MCP (Model Context Protocol): An open standard for connecting AI models to external tools and data sources, enabling AI to interact with any system through a unified protocol.
04

Cowork Flows

Break down a complex task with Claude
Select a task and click the button to see Claude's breakdown.

"The most effective way to use Claude on complex work isn't to hand it a task — it's to think alongside it."

Cowork flows are structured workflows where you and Claude collaborate on complex tasks. The pattern: (1) Break down the task together. (2) Claude drafts; you review and redirect. (3) Iterate section by section rather than end-to-end. (4) Use Claude to check Claude — ask it to critique its own output. This approach produces far better results than one-shot "write me a 50-page report."

Cowork flow: A structured human-AI collaboration pattern where complex tasks are broken into iterative steps, with human review and redirection at each stage.
05

Advanced Tricks

Click any card to flip and reveal the trick
01
Long Document Analysis
tap to flip
Long Document Analysis
Paste large docs in full — Claude handles 200k+ tokens. Ask it to extract, summarize, or cross-reference sections.
"Here is a 40-page report. Find all risk factors and summarize each."
02
XML Tagging
tap to flip
XML Tagging
Wrap prompt parts in <document>, <task>, <examples> tags. Claude parses structured prompts more reliably.
<task>Rewrite this</task> <document>...</document>
03
Artifacts
tap to flip
Artifacts
Use Artifacts for long documents — they persist across turns and can be iteratively edited without re-pasting context.
"Create an artifact for this spec. I'll refine it in steps."
04
Memory via Notes
tap to flip
Memory via Notes
Claude has no persistent memory across chats. Paste a "context note" at the start of each session to simulate memory.
"My context: I'm a PM at a B2B SaaS company building for enterprise..."
05
Structured Output
tap to flip
Structured Output
Specify the exact output format — JSON, table, bullet list — and tell Claude what NOT to include. Constraints outperform instructions.
"Output JSON only. No explanation. No markdown. Fields: title, summary, tags."
Scroll to see all 5 tricks →

"The 20% of Claude tricks that produce 80% of the results — most users never find these."

Five tricks that dramatically improve Claude outputs: (1) XML tagging — wrap different parts of your prompt in <document>, <task>, <examples> tags to give Claude clearer signal. (2) Ask Claude to critique its own output before finalizing. (3) Use artifacts for long documents — they persist and can be iteratively edited. (4) Specify the format in extreme detail. (5) Tell Claude what NOT to do — constraints are often more powerful than instructions.

XML prompting: Using XML tags like <task>, <context>, <examples> in prompts to clearly delineate different parts of the input, improving Claude's ability to follow complex instructions.
06

Claude in Workflows

Click an app to see Claude's workflow
✦
Claude API
📓 Notion
💌 Gmail
💬 Slack
🐙 GitHub
Select an app above
Click any connected app to see how Claude fits into its workflow.

"Claude isn't just a chat interface — it's an intelligence layer you can wire into any process."

Claude's API enables sophisticated integrations. Connect it to Notion to auto-summarize meeting notes. Wire it to Gmail to draft responses in your tone. Link it to GitHub to auto-review pull requests. Use Zapier or Make as the glue between Claude and your existing tools. The pattern is always the same: trigger → Claude does something smart → result goes somewhere useful.

Claude API integration: Using Claude as a processing layer within automated workflows — receiving inputs from apps, generating intelligent outputs, and routing results to downstream systems.

You've mastered Claude Tips!

You now know the models, skills, MCP, cowork flows, and advanced tricks that power pros use every day.

Continue: Build Flows →