Curated · March 2026

Learn Claude

One page to learn everything Claude can do. Courses, tools, tutorials, and people, structured by skill level and linked to primary sources.

Built because the best resources exist. They're just spread across too many platforms.

140+Curated resources
60+Experts & voices
7Steps to Claude fluency
3Skill levels
FreeEvery resource

Nobody prepared you for this.

Me neither.

I'm Sara. Founder, engineer, builder. I dig into AI and tech every day in my own companies. I spend the hours most founders don't have.

    Only when it's worth your time.

    Start at Anthropic Academy →

    What you will find here

    Eleven sections, one reading order. Start at any level that fits your experience. Every entry links to its original source.

    LEARN LEVEL 1 Use Claude LEVEL 2 Build with Claude LEVEL 3 Design Agents Courses, docs, guides from official sources UNDERSTAND Prompt Engineering Context Engineering Agent Environments EXPLORE Tools & Platforms PRACTICE Claude Code Cowork Skills 7 Steps to Fluency People & Platforms

    About Claude

    Claude is an AI system built by Anthropic, designed around principles of helpfulness, safety, and honesty. It operates as a conversational interface at claude.ai, as an agentic coding tool via Claude Code, and as an API for integration into custom applications.

    Use Claude

    Foundational resources for working with Claude in daily tasks: writing, research, structured thinking, and iterative problem-solving. Suitable for all skill levels.

    Core Features You Should Know

    These features turn Claude from a chatbot into a persistent workspace. Understanding them is the difference between casual use and real productivity.

    Build with Claude

    Resources for integrating Claude into applications and development workflows. Covers the API, Claude Code, tool use, and building structured AI systems.

    CourseAnthropic Academy
    Anthropic API Fundamentals
    Structured introduction to the Claude API: endpoint architecture, request formatting, authentication, response handling, and operational best practices.
    FreeCertificateAPI
    CourseAnthropic Academy
    Claude Code in Action
    Pair-programming patterns, automated refactoring, test generation, and Git workflow management. Official certificate.
    FreeCertificateCode
    Docscode.claude.com
    Claude Code: Official Documentation
    Complete reference: installation, CLAUDE.md configuration, MCP server integration, Git workflows, IDE setup, and session management.
    ReferenceTerminal
    Docsdocs.anthropic.com
    Features Overview: Build with Claude
    Technical overview: tool use, file handling, citations, batch processing, structured outputs, streaming, vision, and PDF support.
    CompleteAPI
    TutorialGitHub · Anthropic
    Anthropic Courses (GitHub Repo)
    5 official Jupyter Notebook courses: API Fundamentals, Prompt Engineering, Real World Prompting, Evaluations, Tool Use.
    JupyterHands-on
    TutorialGitHub · Anthropic
    Anthropic Cookbook
    Code recipes for common workflows: tool use patterns, multi-turn conversations, structured outputs, citations, embeddings, RAG, and more.
    Hands-onCookbook
    Guideanthropic.com/engineering
    Context Engineering for AI Agents
    The definitive Anthropic article: compaction, just-in-time retrieval, tool search, sub-agent isolation, CLAUDE.md best practices.
    EssentialEngineering
    Guidedocs.anthropic.com
    Claude 4.x Prompting Best Practices
    Adaptive thinking, extended thinking, vision optimization, context-aware output control, multi-window workflows.
    CurrentClaude 4.x

    Design Agents with Claude

    Advanced resources for building agentic systems: the Agent SDK, Model Context Protocol, reusable skills, multi-agent orchestration, and cloud deployment.

    The augmented LLM: retrieval, tools, memory

    The augmented LLM: the basic building block with retrieval, tools, and memory

    Prompt chaining workflow

    Prompt chaining: sequential steps with programmatic checks between LLM calls

    Routing workflow

    Routing: classify inputs, direct to specialized handlers (Haiku vs. Opus)

    Parallelization workflow

    Parallelization: run sub-tasks simultaneously and aggregate results

    Source: anthropic.com/engineering — Building Effective Agents

    Core Concepts

    The principles that connect everything on this page. How Claude processes information, how you structure it, and how the discipline has evolved from writing prompts to designing complete AI environments.

    Prompt engineering vs context engineering comparison

    The evolution: prompt engineering focuses on writing one instruction. Context engineering manages the full state across turns, tools, memory, and retrieval.

    System prompt calibration: too specific vs too vague

    Finding the right altitude: too specific creates brittleness, too vague lacks guidance. The sweet spot gives strong heuristics without hardcoding logic.

    Source: anthropic.com/engineering — Effective Context Engineering for AI Agents

    Prompts
    Context Windows
    CLAUDE.md & Memory
    Skills & MCP
    Agent Environments
    Foundation
    Prompt Engineering
    Where it starts: writing clear, structured instructions. XML tags for organization, chain-of-thought for reasoning, role-based system prompts for consistent behavior, few-shot examples for calibration.
    The Paradigm Shift
    Context Engineering
    The critical evolution: managing the full context window, not just one prompt. CLAUDE.md for persistent project knowledge. Compaction to summarize and re-start cleanly. Just-in-time retrieval over upfront loading. As context fills past 50%, accuracy degrades.
    The Frontier
    Agent Environments
    The full picture: packaging workflows as reusable Skills. Connecting tools via MCP. Coordinating sub-agents with isolated context windows. Multi-session state with progress files. Designing complete environments where Claude operates autonomously.
    Deep-dive resources

    Tools & Platforms

    Official entry points to the Claude ecosystem. Each serves a distinct function in learning, building, or deploying.

    THE CLAUDE ECOSYSTEM Claude claude.ai · Chat · Projects · Memory Claude Code Terminal · IDE · Agentic coding Cowork Desktop agent · Files · Skills · Sub-agents Integrations Chrome · Excel · PowerPoint · Slack FOUNDATION API & SDK MCP Protocol Skills Connectors Agent SDK Bedrock · Vertex
    Claude.ai
    Chat interface. Projects, Artifacts, Memory, Web Search, file creation.
    claude.ai
    Claude Code
    Agentic coding in terminal, IDE, desktop, and browser. Reads codebases, writes code, runs tests.
    code.claude.com
    Cowork
    Desktop agent for non-coding tasks. File access, multi-step execution, plugins, MCP connectors.
    claude.com/product/cowork
    Claude in Chrome
    Browser agent. Reads pages, navigates sites, fills forms, automates multi-tab workflows.
    anthropic.com
    Claude in Excel
    Spreadsheet agent. Multi-tab workbooks, formula debugging, pivot tables, conditional formatting.
    claude.com/claude-in-excel
    Claude in PowerPoint
    Presentation agent. Template-aware slides, brand compliance, bullets to diagrams.
    claude.com/claude-in-powerpoint
    Claude in Slack
    Team assistant. Summarizes threads, answers from channels, runs Claude Code from chat.
    support.claude.com
    Developer Platform
    API console, Workbench, release notes, billing. The control center for developers.
    console.anthropic.com
    MCP Protocol
    Open standard for tool connectivity. Connect Claude to Drive, Slack, databases, GitHub.
    modelcontextprotocol.io
    Anthropic Academy
    Official learning platform. 11+ courses, certificates, progress tracking.
    anthropic.skilljar.com
    Learn Hub
    Anthropic's landing page for all learning resources. Courses, docs, guides.
    anthropic.com/learn
    Claude Docs
    Complete API documentation. Models, prompt engineering, tools, SDKs, release notes.
    docs.anthropic.com
    Research
    Papers on Constitutional AI, alignment, interpretability. The science behind Claude.
    anthropic.com/research
    DeepLearning.AI × Anthropic
    All Anthropic courses on DeepLearning.AI: Computer Use, Claude Code, MCP, Agent Skills.
    deeplearning.ai

    3 ways to work with Claude

    Before diving into Cowork or Claude Code, understand which mode fits your task. Chat for thinking. Cowork for working. Code for building.

    Thinking
    Chat
    Working
    Cowork
    Building
    Code
    Official Anthropic guide — all three modes explained in detail Read guide →

    Cowork

    Claude's desktop agent for non-coding work. Runs multi-step tasks in sandboxed folders, coordinates sub-agents, and uses built-in skills for documents, spreadsheets, and presentations. macOS and Windows.

    Official & platform resources
    Creator tutorials & playbooks

    Written Guides

    February 2026 · Model Update

    Two new models just landed: Opus 4.6 & Sonnet 4.6

    Sonnet 4.6 (Feb 17) is now the default on claude.ai — Opus-level quality at Sonnet price ($3/$15 per 1M tokens). Opus 4.6 (Feb 5) adds Agent Teams, Claude in PowerPoint, and a 1M token context window. Both include adaptive thinking and major improvements in computer use and coding.

    Sonnet 4.6 announcement → Opus 4.6 announcement → All models overview →

    Claude Code

    Anthropic's agentic coding tool. Runs in terminal, IDE, desktop, and browser. Reads entire codebases, writes and tests code, manages Git, and coordinates sub-agents for complex projects.

    Claude Code agentic loop: Task → Language Model → Gather context, Formulate plan, Take action, Iterate

    How Claude Code processes tasks

    GitHub MCP integration: Task + MCP Server → Claude Code → Read changes, Assess quality, Write report

    MCP integration with GitHub

    Tool use sequence: You → Coding Assistant → Language Model → ReadFile → Response

    Tool use and file operations

    Official & platform resources
    Creator tutorials & playbooks

    Written Guides

    Claude Skills

    Skills are folders with instructions, scripts, and resources that Claude loads dynamically when relevant. They turn Claude from a general assistant into a specialist for specific tasks. Available across Claude.ai, Claude Code, Cowork, and the API. Open standard since December 2025.

    Anatomy of a SKILL.md file: name, description, and context

    Anatomy of a SKILL.md: metadata (name + description) is loaded at startup, body loaded on demand

    Bundling additional files into a skill

    Skills bundle additional files: Claude reads them only when the task requires it

    Progressive disclosure: 3 levels of context loading

    Progressive disclosure: metadata → core instructions → nested resources, loaded level by level

    Skills triggered in context window via system prompt

    How skills activate: system prompt → skill trigger → SKILL.md load → nested file load → task execution

    Code execution via Skills

    Skills can include executable code: Claude runs scripts without loading them into context

    Source: anthropic.com/engineering — Equipping Agents with Agent Skills

    Learn Skills

    7 Steps to Claude Fluency

    A clear sequence from first conversation to production agents. Each step builds on the one before. Click any step to explore resources and outcomes.

    1
    Understand how Claude works
    Use
    After this step: You know what Claude can do, how it processes instructions, and how to structure your first conversations.
    2
    Master prompts and context engineering
    Use
    After this step: You write structured prompts, manage context with Projects and CLAUDE.md, and understand how Claude processes information across sessions.
    3
    Work with Cowork on your desktop
    Use
    After this step: Claude runs multi-step tasks on your computer: organizing files, creating documents, building presentations, extracting data.
    4
    Work with Claude Code
    Build
    After this step: Claude reads your codebase, writes and tests code, manages Git, and runs as your pair-programming partner.
    5
    Build with the API
    Build
    After this step: You integrate Claude programmatically into your own applications: endpoints, authentication, tool use, structured outputs.
    6
    Connect with MCP and Skills
    Agents
    After this step: Claude accesses your data and tools through MCP. You package repeatable workflows as reusable skills and share them across your team.
    7
    Design agent systems for production
    Agents
    After this step: You architect multi-agent systems with the SDK, manage context at scale, and deploy through cloud infrastructure.

    People & Platforms

    Individuals, publications, and communities producing substantive work on Claude, AI systems, and applied workflows.

    Nobody prepared you for this.

    Leading people. Building something from nothing.
    Getting others to believe what you believe.
    Learning before you've mastered what came before.
    Making decisions nobody else can make.

    Me neither.

    I'm Sara. Founder, engineer, builder.
    I dig into AI and tech every day in my own companies.
    I only share what I've built and tested myself.
    I spend the hours most founders don't have.

    Your time doesn't come back.
    I know what it costs to open one more email.
    I make sure it's worth it.

      Only when it's worth your time.

      Know a resource that belongs here? Let me know on LinkedIn.