Everything about Claude - Basics, Artifacts, Skills, Projects, Computer Use, MCP, API, Safety & More
Goal: By the end, you'll master Claude's full ecosystem - from basic chat to building apps, automating tasks, and using advanced tools.
What Claude is, Anthropic, models
Web, mobile, chat, files, vision
Code, docs, diagrams, websites
Knowledge base, custom context
Cowork, computer use, MCP
Claude-specific techniques
API, pricing, integrations
Constitutional AI, limits, roadmap
Claude is an AI assistant built by Anthropic - designed to be helpful, harmless, and honest. It excels at writing, analysis, coding, and long-document understanding.
Conversational AI that understands complex tasks and nuance.
Processes up to 200K+ tokens - entire books or codebases.
Built with Constitutional AI for responsible outputs.
Known for natural, nuanced, high-quality writing.
Strong code generation, debugging, and explanation.
Generates interactive code, diagrams, and documents.
Three tiers designed for different speed, cost, and capability needs.
| Model | Best For | Speed | Context | Cost |
|---|---|---|---|---|
| Claude 3.5 Haiku | Fast, simple tasks | Fastest | 200K | Lowest |
| Claude 3.5 Sonnet | Most tasks, coding, analysis | Fast | 200K | Medium |
| Claude 3 Opus | Complex reasoning, deep analysis | Slower | 200K | Highest |
Most users should default to Claude 3.5 Sonnet - it's the best balance of speed, intelligence, and cost. Use Haiku for bulk/simple tasks, Opus for hardest problems.
Anthropic's unique approach to making AI safe and helpful without massive human annotation.
Learn from vast text data
AI evaluates its own outputs against principles
AI Feedback replaces some human feedback
Reject harmful requests
Still be useful and honest
The constitution includes values from sources like the UN Declaration of Human Rights, Apple's terms of service, and Anthropic's own research. Claude reviews its responses against these principles and improves.
| Feature | Claude | ChatGPT | Gemini |
|---|---|---|---|
| Company | Anthropic | OpenAI | |
| Best For | Writing, long docs, coding, safety | All-rounder, GPTs, DALL-E | Google ecosystem, search |
| Context Window | 200K tokens (best) | 128K | 1M-2M (longest) |
| Artifacts | Yes (unique) | Canvas | No |
| Computer Use | Yes (unique) | No | No |
| Image Generation | No (text only) | DALL-E 3 | Imagen |
| Coding | Excellent | Excellent | Good |
| Writing Quality | Most natural | Good | Good |
claude.ai
Email or Google account
Confirm email & phone
New chat or project
$20/month for more usage
Limited daily messages, access to Claude 3.5 Sonnet
5x more usage, priority access, Projects, Artifacts
Shared workspace, admin controls, higher limits
Recent chats, Projects, favorites, profile.
Your conversation with Claude appears here.
Type prompt, attach files, switch models.
Choose Claude 3.5 Sonnet or Opus.
Side panel for code, documents, previews.
Organize related chats with shared knowledge.
Claude can process a 500-page PDF in one go - competitors often break on documents this large.
Describe what's in an image, identify objects, scenes.
Explain flowcharts, graphs, technical diagrams.
Solve equations from photos of handwritten work.
Extract text from scanned documents and images.
Debug UI issues from screenshots of apps.
Give feedback on designs, wireframes, layouts.
Claude's massive context window is its biggest differentiator. Here's what 200K tokens means in practice:
Upload a full book and ask questions about any chapter.
Upload entire projects for analysis and debugging.
Analyze big spreadsheets without losing context.
Projects let you organize related chats with shared knowledge files and custom instructions.
Artifacts are Claude's most unique and powerful feature - interactive, standalone outputs that open in a dedicated side panel.
React, HTML, Python, JS - runs live.
Markdown, formatted text, reports.
Full HTML pages with CSS & JS.
Charts, illustrations, visual diagrams.
Calculators, games, prototypes.
Flowcharts and sequence diagrams.
Unlike ChatGPT's Canvas (which is editing-focused) or Gemini (no equivalent), Artifacts create standalone runnable apps that Claude and you can iterate on together.
New Project with name
Add a clear description
Add knowledge files
Set custom system prompt
Start new chats within project
Each project can have its own personality. E.g., a "Code Reviewer" project gives strict feedback; a "Brainstorming Partner" project is creative and exploratory.
Skills are modular capabilities that Claude can load on demand - like apps for the AI.
Apply your brand voice and style automatically.
Create editable slide decks via PPTX skill.
Generate spreadsheets with formulas and formatting.
Build your own skills for your workflows.
Call external services and APIs from Claude.
Growing ecosystem of pre-built skills.
When Claude recognizes a task matches a skill, it loads that skill's instructions and executes it - making Claude extensible without retraining.
A groundbreaking capability - Claude can see and operate your computer like a human.
Computer Use is in beta. Always supervise Claude's actions, use it in sandboxed environments, and review its decisions. It can make mistakes on unfamiliar UIs.
Cowork is Claude's agentic desktop feature - Claude works alongside you on your files and tasks.
Currently available for Max and Team plan subscribers on macOS and Windows, with sandboxing for safety.
MCP is an open standard created by Anthropic that lets Claude connect to any tool, database, or service.
One standard for connecting Claude to external systems.
Connect to Google Drive, Notion, Slack, databases.
Trigger workflows, fetch live data, take actions.
Index Slack messages, docs, wikis for Claude to search.
Let Claude run SQL or NoSQL queries safely.
Used by Claude Desktop, Cursor, and many tools.
Design Mode lets you preview and edit visual artifacts directly - UI mockups, SVGs, and layouts.
See rendered HTML/CSS/React before deploying.
Tell Claude to change colors, layout, copy inline.
Test mobile, tablet, desktop breakpoints.
Use shadcn, Tailwind, custom styles.
Make changes and preview in seconds.
Copy the code or publish live.
Claude responds exceptionally well to structured, specific prompts with clear context.
State exactly what you want - no hidden assumptions.
Claude uses 200K tokens - give it everything.
"You are an expert copywriter..."
Specify exact format: bullets, tables, JSON.
Claude is trained to recognize <tag> sections.
Ask Claude to refine its own answer.
Anthropic trained Claude on XML-tagged prompts - using tags dramatically improves clarity and output quality.
Tags separate concerns, make prompts scannable, and Claude has learned to respect them as natural boundaries.
| Model | Input | Output |
|---|---|---|
| Claude 3.5 Haiku | $0.80 | $4.00 |
| Claude 3.5 Sonnet | $3.00 | $15.00 |
| Claude 3 Opus | $15.00 | $75.00 |
Use Haiku for simple classification, Sonnet for most tasks, Opus only when needed.
Cache repeated system prompts - saves up to 90% on costs.
50% discount for async batch processing of large volumes.
Cache large contexts (saves time and money).
Process large jobs at 50% discount.
Native function calling and tools in API.
Send images in messages, get descriptions.
Estimate costs before running requests.
Get responses as they're generated.
Define tools (functions) Claude can call - like giving the AI hands to interact with your systems.
Claude returns a "tool_use" block → your code runs the function → returns result to Claude → Claude gives final answer. Repeat for multi-step tasks.
AI-powered IDE for coding
Docs and knowledge management
Team chat and summaries
Workflow automation
Docs and Sheets
Meeting transcripts and summaries
E-commerce product descriptions
No-code workflows
Claude Code is Anthropic's official CLI tool for AI-assisted development directly in your terminal.
npm install -g @anthropic-ai/claude-code - then run claude in any project directory.
Ask Claude to review your changes before committing.
"Convert all class components to function components."
Generate unit tests for untested modules.
Auto-generate README and JSDoc comments.
Paste error, Claude finds the bug in your codebase.
"Add a search bar to this page" - Claude writes the code.
Blog posts, whitepapers, ebooks with brand voice.
Personalized sequences that feel human.
Platform-specific posts in your tone.
Synthesize reports and surveys into insights.
Personalized cold emails and follow-ups.
Multiple variations for A/B testing.
Anthropic's approach to building AI that's helpful, harmless, and honest.
No real-time internet access in API.
Cannot create images - vision only.
Text-based only.
Can make calculation errors.
May decline borderline requests that aren't harmful.
Each conversation starts fresh (unless via Projects).
Paid tier conversations aren't used to train models.
Free chats kept 30 days for safety; paid for abuse prevention.
Zero data retention, custom terms, SSO available.
Process data in specific regions (US, EU, etc.)
Track team usage and access (Team plan+)
API keys, PII, medical records, secrets
| Tool | Best For | Unique Strength |
|---|---|---|
| Claude | Writing, code, long docs, safety | Artifacts, Computer Use, 200K context |
| ChatGPT | All-rounder, image gen | DALL-E, GPTs, plugins |
| Gemini | Google ecosystem, search | 2M context, multimodal |
| Perplexity | Research with sources | Search-first with citations |
| Copilot | Microsoft 365 users | Office integration |
| Mistral | Open-source deployments | Self-hostable, European |
Use Claude for writing, long docs, code review, and Artifacts. Add ChatGPT for image gen, Gemini for Google tasks, Perplexity for research.
| Plan | Price | Best For |
|---|---|---|
| Free | $0 | Trying out Claude with limited daily messages |
| Pro | $20/month | Individuals who use Claude heavily for work |
| Max | $100-$200/month | Power users needing 5x-20x more usage |
| Team | $25/user/month | Small teams with collaboration needs |
| Enterprise | Custom | Large orgs with security and compliance needs |
| API | Pay per token | Developers building applications |
Better long-running autonomous workflows.
Video understanding coming.
Continued speed and efficiency gains.
Cross-session context and personalization.
Expanding language support.
Deeper integrations and compliance.
Comfortable with chat, files, Artifacts
Built real apps and workflows with Claude
"What could be better about your last response?"
Combine expert personas for richer answers.
For complex tasks, ask Claude to plan first then execute.
Ask for citations when facts matter.
Use prompt caching for repeated system prompts.
Compare Sonnet vs Opus for hard problems.
Notion uses Claude to summarize meetings and answer questions across millions of docs.
Cursor built their entire AI code editor experience with Claude.
Lexis+ AI uses Claude to help lawyers research case law and draft documents.
Scientists at Genentech use Claude to analyze research papers and accelerate drug discovery.
Companies use Claude-powered chatbots to handle tier-1 support with human-quality answers.
Bridgewater uses Claude for economic analysis and decision-making support.
Claude handling complex multi-step work autonomously.
MCP becomes standard across all AI tools.
Claude runs continuously alongside you.
Claude learns your style and preferences deeply.
Better multilingual support and regional understanding.
Constitutional AI principles across the industry.
From basic chat to Artifacts, Skills, MCP, and API - you have everything to use Claude like a professional across any task.
XML tags, Projects, long context
Interactive code, docs, designs
MCP, Computer Use, Skills, Cowork
Claude Code CLI and API
Constitutional AI principles
Real apps and workflows
🎬 If this helped, like, share, and subscribe!
💬 What will you build with Claude? Comment below.