ClaudeClaude / CLI

Claude Code

Claude Code is Anthropic's agentic coding tool — a command-line program that gives Claude direct access to a real development environment: reading and editing files, running shell commands, searching a codebase, and iterating until a task is done. This guide covers it completely, topic by topic, from first install to designing a workflow around it.

Fundamentals

What it is, getting it installed, and the two things that load every session.

What Is Claude Code?What Claude Code actually is, how it differs from a chat assistant, and where it fits among Claude's other surfaces.3 min read

Installation & SetupInstalling Claude Code, authenticating, and what happens the first time you run it on a project.3 min read

Project Memory: CLAUDE.md & AGENTS.mdHow CLAUDE.md and AGENTS.md work, what belongs in them, and how to structure them for larger projects.3 min read

Permission ModesHow Claude Code's permission system works, the different modes available, and how to tune it per command or tool.3 min read

Core Capabilities

The tools behind everything it does: files, the shell, search, and git.

Reading & Editing FilesHow file reading and editing actually works under the hood, and why targeted edits matter more than full-file rewrites.3 min read

Running Shell CommandsHow the bash tool works, what it's used for, and the safety considerations that come with letting an agent run real commands.3 min read

Searching the CodebaseHow glob and grep-style search let Claude Code work in large, unfamiliar codebases without loading everything into context.2 min read

Git-Native WorkflowsHow Claude Code handles git: inspecting history, writing commits from a real diff, branching, and opening pull requests.3 min read

Web Search & FetchLooking up current documentation and fetching specific URLs, and when this matters more than relying on training data.2 min read

SubagentsWhat subagents are, why they exist, and when delegating a chunk of work to one beats doing everything in the main conversation.3 min read

Advanced Features

The configurable layer: plans, commands, skills, hooks, MCP, and background work.

Plan ModeGetting Claude to lay out and get sign-off on an approach before any file is touched, and when that step is worth the friction.2 min read

Slash CommandsBuilt-in and custom slash commands: packaging a repeatable prompt into a single, named, reusable invocation.2 min read

SkillsHow skills bundle task-specific knowledge that loads on demand, and how that differs from always-loaded project memory.3 min read

HooksRunning your own shell commands automatically around Claude Code's lifecycle events, for formatting, enforcement, and logging.3 min read

MCP: Model Context ProtocolConnecting Claude Code to external tools and data sources through a standard protocol instead of one-off integrations.3 min read

Background TasksRunning long commands in the background so a session can keep working instead of blocking on a slow build or test run.3 min read

Putting It Together

A real workflow end to end, and the habits that make it consistently effective.

A Typical Workflow, Step by StepA realistic walkthrough of how a real task moves from a plain-language request to a verified, shippable change.3 min read

Best PracticesThe habits that make Claude Code consistently effective across many tasks, beyond what any single feature covers on its own.3 min read

Curious what this looks like in practice? This entire site was built with Claude Code — read how ConsLabs itself was built.