wesammustafa/Claude-Code-Everything-You-Need-to-Know

Claude Code – Everything You Need to Know

What it is – A community‑maintained, step‑by‑step guide for Claude Code, Anthropic’s official command‑line interface that lets you run Claude (the large‑language‑model) directly from your terminal. The guide walks you from the first installation through prompt engineering, custom Skills (slash commands), Hooks, sub‑agents, dynamic workflows, and the Model‑Context‑Protocol (MCP) for external tool integration.

Who it’s for

• New users who want a quick, 15‑minute path to a working Claude Code session.
• Experienced users looking for deeper coverage of Skills, Hooks, MCP, and multi‑agent setups.
• Team leads / automation engineers interested in building reusable agent teams, dynamic JavaScript workflows, and the BMAD framework for AI‑agent orchestration.

Key sections

Section	What you’ll learn
Setup	One‑liner install (`curl …
Prompt Engineering	Structured “Explore → Plan → Code → Commit” workflow, effort‑level dial (low → max + ultracode), and how to use ultrathink for deeper reasoning.
Skills / Slash Commands	How to create reusable markdown‑based commands (.claude/commands/*.md) and the built‑in skill set shipped with the repo (PR hygiene, TDD, UX, etc.).
Hooks	Automatic Python hooks that run on session start, permission changes, or when Claude needs user input.
Subagents & Agent Teams	Defining isolated agents (.claude/agents/*.md), using the experimental Agent Teams feature, and a library of ready‑made role prompts.
Dynamic Workflows	JavaScript workflow scripts (.claude/workflows/*.js) that coordinate multiple agents, e.g., a stale‑docs audit.
MCP (Model‑Context‑Protocol)	Configuring Claude to call external browsers, databases, or APIs from within a session.
Productivity frameworks	Effort levels, Fast Mode, the “Super Claude” framework, and the BMAD method for building AI‑agent pipelines.

Repository layout (the parts you can copy into your own project)

• .claude/commands/ – 7 ready‑to‑use slash skills (e.g., /pr, /review).
• .claude/skills/ – an example Agent Skill (/claude‑md‑review).
• .claude/agents/ – 5 starter sub‑agents plus 10 specialized role prompts.
• .claude/workflows/ – a sample dynamic workflow (/stale‑docs‑audit).
• .claude/hooks/ – a Python notification hook (requires the uv Python installer).
• .claude/settings.json – default permission modes and hook wiring.

How to get started

# Install Claude Code (macOS/Linux/WSL)
curl -fsSL https://claude.ai/install.sh | bash
# Authenticate (opens browser)
claude
# Launch in any repo
cd my‑project && claude
# Try a built‑in command
/explain what this codebase does
# Or run a custom skill from the repo
/pr

If you want to reuse the example configuration, clone the repo and copy only the files you need, reviewing each for security before execution.

Why it matters – Claude Code is the primary way developers interact with Anthropic’s Claude models from the terminal, giving the model real access to the repository file tree, test runners, and external tools. This guide demystifies the five extension points (Skills, Hooks, Subagents, Workflows, MCP) and provides concrete, copy‑and‑paste examples, turning a generic CLI into a programmable AI‑assistant platform.

License & maintenance – MIT‑style open source (see LICENSE). The README notes a review in July 2026 and a correction in Oct 2026, indicating active upkeep.