Claude Code Guide
30 pages covering every Claude Code feature — from install to CI/CD automation.
Getting Started
Install, configure, and get productive in your first session.
CLAUDE.md
Learn everything about CLAUDE.md — the file that gives Claude Code persistent memory about your project. Covers structure, @imports, best practices, and real templates for monorepos.
First Session
What to expect in your first Claude Code session, how to give effective context, common mistakes to avoid, and how to set up CLAUDE.md for 10x productivity from day one.
Headless Mode
Run Claude Code non-interactively in CI/CD pipelines, shell scripts, and automation. Covers the --print flag, piping input, environment setup, and GitHub Actions integration.
Install
Step-by-step guide to installing Claude Code on macOS, Linux, and Windows in 2026. Covers npm global install, authentication with Anthropic API keys, and verifying your setup.
Permission Modes
Understand Claude Code's four permission modes — auto, plan, dontAsk, and bypassPermissions — when to use each, security implications, and how to configure them for your team.
settings.json
Complete reference for Claude Code's settings.json file. Covers all fields, permission modes, environment variables, MCP server config, and the difference between project and user-level settings.
Skills
Write reusable skills that extend Claude Code for your project.
Best Skills 2026
The 50 highest-value Claude Code skills in 2026, organized by category. Each skill includes its purpose, when to use it, and a copy-paste template to get started immediately.
SKILL.md Format
The complete format reference for Claude Code skill files. Learn every section, when to use @imports, how to write effective instructions, and how to test skill quality.
Skills vs Commands vs Hooks
Understand the difference between Claude Code Skills, slash commands, and hooks. A practical decision framework for choosing the right mechanism for each automation need.
What Are Skills
Learn what Claude Code Skills are, how they differ from slash commands, how Claude discovers and invokes them, and why they are the fastest way to encode team workflows into Claude.
Write a Skill
Learn the SKILL.md format for writing Claude Code skills. Includes templates for common skill types, examples of effective vs ineffective skills, and tips for making skills reusable across the team.
Hooks
Automate reactions to Claude Code events with shell scripts.
Hooks Overview
Complete guide to all 5 Claude Code hook types — PreToolUse, PostToolUse, UserPromptSubmit, SessionStart, and SessionStop. Learn when each fires, how to configure them, and real-world examples.
PostToolUse Hook
Learn how to write PostToolUse hooks in Claude Code to auto-format files after edits, run tests after writes, log outputs, and trigger side effects after every tool call completes.
PreToolUse Hook
Learn how to write PreToolUse hooks in Claude Code to validate tool arguments, block dangerous operations, enforce policies, and log every tool call before it executes.
SessionStart Hook
The SessionStart hook runs once when Claude Code begins a new session, ideal for loading environment state, fetching fresh context, or logging session metadata.
UserPromptSubmit Hook
The UserPromptSubmit hook fires before Claude processes your message, letting you validate input, inject context, or block disallowed queries automatically.
MCP Servers
Install and build MCP servers to give Claude new capabilities.
Best MCP Servers
The best MCP servers for Claude Code covering databases, browsers, GitHub, file systems, search, and more — with install commands for each.
Build an MCP Server
Build a custom MCP server in TypeScript or Python to give Claude Code new tools — database queries, internal APIs, or any custom action your workflow needs.
MCP Overview
MCP (Model Context Protocol) servers extend Claude Code with new tools — databases, APIs, browsers, file systems — without writing Claude-specific code.
MCP vs Skills
MCP servers add executable tools to Claude Code; skills add reusable prompt instructions. Understand the difference to architect your Claude Code setup correctly.
Power User
Background tasks, worktrees, CI/CD, and parallel agents.
Background Tasks
Run Claude Code as a background process to execute long-running tasks, parallel workstreams, or automated jobs while you continue other work.
Context Management
Manage Claude Code's context window to avoid hitting limits on long sessions — use /compact, CLAUDE.md references, and smart file reading strategies.
GitHub Actions
Run Claude Code in GitHub Actions to automate PR reviews, generate changelogs, fix failing tests, and enforce code standards on every commit.
Parallel Agents
Run multiple Claude Code instances in parallel git worktrees to complete independent tasks simultaneously — reducing hours of sequential work to minutes.
Remote Triggers
Trigger Claude Code tasks remotely via webhooks, GitHub Actions, scheduled cron jobs, or API calls — turning Claude into an automated background agent.
Git Worktrees
Use git worktrees with Claude Code to run multiple Claude tasks on separate branches simultaneously without stashing, context-switching, or merge conflicts.
Comparisons
How Claude Code stacks up against Cursor, Aider, and Codex CLI.
Model Selection
Choose the right Claude model for Claude Code tasks — Sonnet for everyday coding, Opus for complex architecture and hard bugs, Haiku for fast cheap automation.
vs Aider
Claude Code vs Aider — both are terminal AI coding tools, but they differ in architecture, model support, workflow automation, and extensibility.
vs Codex CLI
Claude Code vs OpenAI Codex CLI — two terminal AI coding agents from competing AI labs. Compare features, model quality, tool execution, and extensibility.
vs Cursor
Claude Code vs Cursor — a detailed comparison of features, pricing, model flexibility, context handling, and use cases for developers choosing between the two.