Claude Skills Guide

Claude Code Getting Started: Complete Beginner Guide

Claude Code is Anthropic’s command-line AI coding tool — a terminal-native assistant that reads your files, writes code, runs commands, and works through development problems alongside you. This hub is the single starting point for every beginner. It covers what Claude Code is, how to install it, what skills are, how to run your first session, and links to every getting-started article in the library.

Table of Contents

  1. What Is Claude Code
  2. Installation Guide
  3. Understanding Skills
  4. Your First Session
  5. Essential Skills for Beginners
  6. Understanding Auto-Invocation
  7. Security and Permissions
  8. Quick Reference Table
  9. Complete Article Index

What Is Claude Code

Claude Code is not a chatbot. It is Anthropic’s AI coding assistant that runs in your terminal, directly inside your project. You launch it with claude, and it immediately has access to your file tree, your shell, and your development environment. It can read any file you give it permission to see, write and modify files, execute shell commands, and check their output — all within one interactive session.

The key distinction from web-based AI tools is integration. When you describe a change in Claude Code, Claude does not hand you a snippet to paste somewhere. It reads the actual file, makes the actual change, runs your test suite to confirm nothing broke, and reports back. The full loop — read, plan, write, verify — happens without you leaving the terminal.

What makes Claude Code genuinely different from other AI coding tools is the skills system. Skills are Markdown files that extend Claude’s default behavior with domain-specific expertise. Instead of being a general assistant, Claude becomes a specialist for the exact task at hand: enforcing test-driven development, processing spreadsheets, generating design-system-compliant components, or maintaining session memory across days of work. Skills are composable, versioned, and shareable.

Claude Code works on macOS, Linux, and Windows (through WSL). It supports any editor, any stack, and any build system. There is no IDE plugin to install, no browser tab to switch to.

Start here: What Is Claude Code and Why Developers Love It in 2026 — Claude Skills Explained Simply for Non-Programmers — Is Claude Code Worth It? An Honest Beginner Review 2026

Installation Guide

Getting Claude Code running takes about five minutes. You need Node.js 18 or later, an Anthropic API key, and access to a terminal.

Step 1 — Install via npm:

npm install -g @anthropic-ai/claude-code

Verify the installation succeeded:

claude --version

If you see “command not found” after installing, close and reopen your terminal. On some systems you may need to add your global npm bin directory to your PATH.

Step 2 — Authenticate:

Get an API key from console.anthropic.com under “API Keys”, then set it as an environment variable:

export ANTHROPIC_API_KEY=your-key-here

Add this to your .bashrc or .zshrc to make it permanent. The key is stored locally — you authenticate once and it persists.

Step 3 — Navigate to a project:

cd ~/your-project
claude

Claude Code opens an interactive session scoped to that directory. Claude can see everything inside the directory and its subdirectories.

Platform-specific setup: If you are on Windows, run Claude Code inside WSL2. See Claude Code Skills in WSL2: A Practical Setup Guide for a complete walkthrough. Docker users can also run Claude Code containerized — see Claude Code with Docker: Container Setup Guide.

Keeping skills in sync across machines: Claude Code Dotfiles Management and Skill Sync Workflow covers how to keep your skill files in a dotfiles repo so they follow you everywhere.

More installation and setup guides:

Understanding Skills

Skills are the feature that separates Claude Code from every other AI coding tool. A skill is a plain Markdown file stored in ~/.claude/skills/. It has two parts: a YAML front matter block at the top that holds metadata, and a Markdown body below that contains the actual instructions Claude follows when the skill is active.

Here is a minimal skill file:

---
name: my-skill
description: "What this skill does in one sentence"
---

You are a specialist for this type of work.

When invoked, always:
1. Do the first thing
2. Do the second thing
3. Report results clearly

The name field is how you invoke the skill manually: /my-skill. The description field is used by Claude to understand when a skill is relevant and when to activate it automatically.

Claude Code ships with a set of official built-in skills covering the most common workflows: pdf for document processing, tdd for test-driven development, xlsx for spreadsheet automation, docx for Word documents, pptx for presentations, frontend-design for UI components, and supermemory for persistent context across sessions.

Beyond official skills, there is a growing ecosystem of community-built skills for specific stacks, workflows, and domains. You can download them manually or contribute your own.

Key things to understand about skills:

Skills are loaded lazily. Claude reads the metadata header of every skill file at startup, but loads the full body only when a skill is invoked. This keeps sessions fast even when you have dozens of skills installed.

Skills are composable. You can combine two skills in a single session — the tdd skill and the pdf skill can both be active at once if your task involves generating a test report as a PDF.

Skills are scoped. A skill’s instructions apply only while the skill is active in the current session. Different projects can use different skill configurations.

Go deeper:

Your First Session

Once Claude Code is installed and authenticated, here is a practical walkthrough for your first real session.

Open Claude Code in your project:

cd ~/your-project
claude

Claude confirms the project directory it has scoped to and is ready for input.

Try a simple exploration task first:

Ask Claude to describe the project structure:

Summarize this codebase: what it does, the main entry points, and any areas that look like they need attention.

Claude reads your files and responds with a structured summary. This is a good way to verify it has access to your code and that the session is working.

Make a concrete change:

Ask for something specific:

Add input validation to the user registration endpoint. Check that email is a valid format and password is at least 8 characters. Return a 400 with a descriptive error message if validation fails.

Claude reads the relevant file, writes the validation logic, and may run your test suite to confirm the change does not break existing tests.

Invoke your first skill:

Load the TDD skill and ask for tests:

/tdd write unit tests for the validation logic you just added

Claude generates test cases covering the happy path, invalid email format, and short password edge cases — following test-driven development conventions enforced by the skill.

End the session:

Type /exit or press Ctrl+C.

Tutorials for your first session:

Essential Skills for Beginners

These are the skills worth understanding first. Each covers a common development workflow and demonstrates how skills change what Claude Code can do.

PDF Skill

The pdf skill is the best demonstration of what a skill actually does. Without it, Claude can discuss PDF content you paste into the chat. With it, Claude becomes a full document processing engine — extracting tables, reading multi-column layouts, filling forms, and generating new PDF documents from structured data.

Invoke it explicitly:

/pdf extract all tables from Q3-financial-report.pdf as markdown

/pdf merge invoices/*.pdf into one document ordered by date

The pdf skill handles scanned documents (with OCR), form fields, and multi-page layouts. For developers processing contracts, reports, or specifications in bulk, it replaces entire manual workflows.

Read more: Best Claude Code Skills to Install First in 2026

TDD Skill

The tdd skill enforces test-driven development: write a failing test first, write the minimum implementation to pass it, refactor. It works with pytest, Jest, Vitest, and Bun Test.

/tdd write pytest tests for the authentication module

/tdd given this failing Jest test, implement UserService.authenticate() to make it pass

The skill does not just generate tests — it helps you think through edge cases you would miss on your own: expired sessions, concurrent logins, malformed inputs. For any developer building anything that needs to work reliably, TDD skill is the one to add first.

Read more: Claude TDD Skill: Test-Driven Development Guide (2026) — Automated Testing Pipeline with Claude TDD Skill (2026)

Frontend Design Skill

The frontend-design skill specializes in building UI components that match a design system. It understands responsive layout, accessibility requirements (ARIA attributes, semantic HTML), and CSS design tokens. It works with React, Vue, Svelte, and vanilla JavaScript.

/frontend-design create a responsive navigation header with logo and mobile hamburger menu, following the design tokens in src/styles/tokens.css

The skill does not just generate generic markup — it reads your actual design tokens and applies them, keeping output consistent with your existing system.

Read more: Claude Frontend Design Skill Review and Tutorial — Best Claude Code Skills for Frontend Development

XLSX Skill

The xlsx skill automates spreadsheet work: reading data from Excel files, generating new sheets, building formulas, and transforming data between formats.

/xlsx read sales-data.xlsx and generate a pivot table by region and product category

/xlsx create an expense report template with SUM formulas and conditional formatting for over-budget rows

For developers or analysts who spend time manually working with spreadsheets, the xlsx skill eliminates the most repetitive parts of that work.

Read more: Claude /xlsx Skill: Spreadsheet Automation Guide

SuperMemory Skill

The supermemory skill gives Claude persistent memory across sessions. By default, Claude Code starts fresh each session with no recollection of previous work. SuperMemory writes structured notes to disk and reloads them at session start, giving Claude context about your project, your preferences, and previous decisions.

/supermemory remember that we use the repository pattern in the services layer and all error handling goes through the ErrorService class

On your next session, Claude already knows this. You spend less time re-explaining context and more time building.

Read more: Claude SuperMemory Skill: Persistent Context Guide 2026

More essential skill guides:

Understanding Auto-Invocation

Claude Code skills can be invoked in two ways: explicitly with a slash command (/skill-name), or automatically when Claude detects that the context matches a skill’s trigger conditions.

Auto-invocation is controlled by the triggers block in a skill’s front matter. When you send a message, Claude checks whether any installed skill has a trigger phrase that matches what you are asking. If there is a match, Claude loads that skill’s body and applies its instructions without you typing anything extra.

For example, a skill whose description mentions PDF extraction will activate automatically when you ask “can you read this PDF file and pull out the invoice totals?” You never type /pdf. Claude detects the match and loads the skill.

Auto-invocation keeps sessions fast for common tasks. For less common skills, or when you want precision control over which skill is active, explicit invocation with the slash command is more reliable.

Important: auto-invocation only works when the skill is installed in ~/.claude/skills/. If a skill is not installed, no trigger matching happens for it.

When auto-invocation fails: there are several common reasons a skill does not fire automatically — trigger phrases that are too narrow, competing triggers across multiple skills, or context that does not strongly match. The troubleshooting guides below cover each failure mode.

Auto-invocation guides:

Security and Permissions

Claude Code is an agentic tool. It can read files, execute shell commands, write code, and make network requests. Understanding the permissions model is important before using it on real projects.

The layered architecture:

Claude Code permissions work in layers:

Session-level settings
  └── Skill-level overrides
        └── Hook-level enforcement
              └── Tool-level capabilities

Each layer can only restrict — never expand — the permissions of the layer above it. A skill cannot grant itself access to tools the session has disabled.

Default tool access: By default, all built-in tools are available — Read, Write, Bash, WebFetch, WebSearch, and Glob. This is intentional for usability. You restrict access via .claude/settings.json.

Restricting tool access:

{
  "permissions": {
    "allow": ["Read(**)", "Glob(**)", "Bash(git *)"],
    "deny": ["WebFetch(*)", "WebSearch(*)"]
  }
}

This configuration allows Claude to read files and run git commands only, with no network access.

Hooks for enforcement: Claude Code’s hooks system lets you intercept any tool call before it executes. You can log, inspect, modify, or block any action — giving you audit logs and hard guardrails without having to trust Claude’s judgment entirely.

Common permission errors and fixes:

Quick Reference Table

Topic Article Difficulty
What Claude Code is What Is Claude Code and Why Developers Love It Beginner
Plain-language skills intro Claude Skills Explained Simply for Non-Programmers Beginner
Full beginner walkthrough Claude Code for Beginners: Getting Started 2026 Beginner
Honest pros/cons review Is Claude Code Worth It? An Honest Beginner Review Beginner
Skills to install first Best Claude Code Skills to Install First in 2026 Beginner
Where to find skills Claude Skills Directory: Where to Find Skills Beginner
Skill file format Skill MD File Format Explained With Examples Intermediate
Write your first skill How to Write a Skill MD File for Claude Code Intermediate
Full skill specification Claude Skill .md File Format: Full Specification Guide Intermediate
Auto-invocation explained Claude Skills Auto Invocation: How It Works Intermediate
Permissions and security Claude Code Permissions Model and Security Guide Intermediate
TDD skill guide Claude TDD Skill: Test-Driven Development Guide Intermediate
Frontend design skill Claude Frontend Design Skill Review and Tutorial Intermediate
Spreadsheet automation Claude /xlsx Skill: Spreadsheet Automation Guide Intermediate
Persistent memory Claude SuperMemory Skill: Persistent Context Guide Intermediate
Skill not triggering Claude Skill Not Triggering: Troubleshoot Guide Intermediate
YAML parsing errors Claude Skill YAML Front Matter Parsing Error Fix Intermediate
Token optimization Claude Skills Token Optimization: Reduce API Costs Intermediate
Sharing skills with a team How to Share Claude Skills with Your Team Intermediate
WSL2 setup Claude Code Skills in WSL2: A Practical Setup Guide Intermediate
Docker integration Claude Code with Docker: Container Setup Guide Intermediate
Context window management Claude Skills Context Window Management Best Practices Intermediate
Skill versioning Claude Skill Versioning: Semver Best Practices Advanced
Enterprise skill structure Structuring Claude Skills for Large Enterprise Codebases Advanced
Multi-agent error recovery Claude Code Multi-Agent Error Recovery Strategies Advanced
Production AI agents Building Production AI Agents with Claude Skills in 2026 Advanced

Complete Article Index

The articles below cover every getting-started topic in the library, organized by sub-topic. Use this as a complete map for any question about Claude Code fundamentals.

What Is Claude Code

Articles explaining the platform, its purpose, and why developers choose it.

Installation and Environment Setup

Getting Claude Code running on different platforms and configurations.

Skill File Format and Structure

Understanding and writing skill.md files from first principles.

Discovering and Managing Skills

Finding, installing, and organizing skills across projects.

Auto-Invocation

How auto-invocation works, how to configure it, and how to fix it when it does not.

Security and Permissions

Understanding and configuring Claude Code’s permission model.

Troubleshooting Errors

Fixing the most common errors you will encounter with Claude Code skills.

Performance, Tokens, and Optimization

Making Claude Code faster and more cost-efficient.

Memory and Context

Persistent memory, session state, and context architecture.

Skills by Developer Role

Role-specific skill recommendations and configurations.

Skills by Technology Stack

Stack-specific skill guides for common platforms and languages.

Integrations and Automation

Connecting Claude Code skills to external services and workflows.

Automation Workflows

Complete workflow automation guides using Claude Code skills.

Building Projects with Claude Code

Step-by-step project guides that demonstrate skills in action.

“What Is the Best Skill for…” Questions

Answers to specific skill selection questions.

“How Do I…” Practical Questions

Concise answers to common operational questions.

“Why Does…” Diagnostic Questions

Explaining unexpected Claude Code behavior.

Industry and Domain-Specific Skills

Guides for specific industries and regulated domains.

Where to Go Next

Once you have the basics down, the rest of the library is organized by topic:

*Built by theluckystrike — More at zovo.one *

Explore Our Developer Guides

© 2026 theluckystrike Developer Guides