Claude Code is an AI-powered coding assistant by Anthropic that lives in your terminal, VS Code, or browser. It reads your entire codebase, writes code, fixes bugs, runs tests, creates git commits and pull requests — all by understanding plain English instructions.

How do I install Claude Code?

On macOS or Linux: curl -fsSL https://claude.ai/install.sh | bash. On Windows PowerShell: irm https://claude.ai/install.ps1 | iex. You can also install via Homebrew, NPM (npm install -g @anthropic-ai/claude-code), or use it in your browser at claude.ai/code with no install required.

Is Claude Code free to use?

Claude Code requires a Claude.ai account (recommended) or Anthropic Console account. There are free and paid subscription options at claude.ai. The web version at claude.ai/code requires a paid subscription.

How do I use Claude Code to fix bugs?

Paste the error message directly into Claude Code and describe the symptom. For example: 'I'm seeing this error when I run npm test: [paste error]'. Claude will trace through your codebase, identify the root cause, and implement a fix. Always ask it to explain what caused the bug so you learn from it.

What is CLAUDE.md and how do I use it?

CLAUDE.md is a Markdown file you create in your project root that Claude Code reads at the start of every session. Use it to store coding standards, architecture decisions, build commands, and team conventions. Create it automatically with the /init command inside Claude Code.

Can Claude Code connect to GitHub, Jira, or Slack?

Yes. Claude Code uses the Model Context Protocol (MCP) to connect to hundreds of external tools including GitHub, Sentry, Slack, Figma, Jira, Postgres databases, and more. Add a server with: claude mcp add --transport http github https://api.githubcopilot.com/mcp/

How do I use Claude Code in GitHub Actions?

Run /install-github-app inside Claude Code and follow the prompts. Once set up, tag @claude in any GitHub PR or issue comment and Claude will automatically analyze the code, implement fixes, and open pull requests.

What are Claude Code subagents?

Subagents are specialized AI personalities that Claude Code can delegate tasks to. Built-in subagents include Explore (read-only fast search), General-Purpose (full access), and Plan (analysis for plan mode). You can create custom subagents like code-reviewer or debugger and store them in .claude/agents/ as Markdown files.

Complete Practical Guide — Based on Official Docs

Master
Claude Code

Everything in the official docs — turned into hands-on practical knowledge with real commands, real examples, and zero fluff. Made for students and junior devs starting from scratch.

8

Core Sections

40+

Real Commands

15+

Live Workflows

5 min

To Install

Start Installing → Jump to Cheatsheet

// step 0 — prerequisites

Installation & Setup

You need a Claude.ai or Anthropic Console account. Then pick your OS below. The entire process takes under 5 minutes.

⚠️

Before You Begin

Sign up at claude.ai (recommended for individuals) or console.anthropic.com (for API/teams). You'll log in on first run — credentials are stored automatically after that.

Recommended. Auto-updates in the background. Works on any Mac or Linux terminal.

Terminal — bash/zsh

# 1. Install with one command

$ curl -fsSL https://claude.ai/install.sh | bash

Downloading Claude Code...

Installing to /usr/local/bin/claude...

✓ Claude Code installed successfully!

# 2. Navigate to your project

$ cd ~/projects/my-app

# 3. Start Claude Code (logs in on first run)

$ claude

→ Opening browser for login...

✓ Logged in! Claude Code is ready.

>

Requires Git for Windows — download it first from git-scm.com/downloads/win

PowerShell (Run as Administrator)

# Option A — PowerShell

PS> irm https://claude.ai/install.ps1 | iex

# Option B — CMD

> curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

# Option C — WinGet

PS> winget install Anthropic.ClaudeCode

# Note: WinGet/Homebrew don't auto-update. Run upgrade periodically.

Homebrew

$ brew install --cask claude-code

# Upgrade manually when needed:

$ brew upgrade claude-code

⚠ Homebrew does NOT auto-update. Check periodically!

Requires Node.js 18 or newer. Check with node --version

npm

# Install globally

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

# Verify install

$ claude --version

claude-code 1.x.x

Prefer a GUI? Install the VS Code extension — no terminal needed to get started.

1

Open Extensions Panel

Press Cmd+Shift+X (Mac) or Ctrl+Shift+X (Windows/Linux)

2

Search "Claude Code"

Find the extension by Anthropic. Click Install. Also works in Cursor IDE.

3

Open Claude Code

Press Cmd+Shift+P, type "Claude Code", select "Open in New Tab".

Zero install. Open your browser and go. Great for quick tasks or working on repos you don't have locally.

🌐

Web Access

Go to claude.ai/code in any desktop browser or the Claude iOS app. Requires a paid Claude.ai subscription. Start async tasks and come back when done.

// step 1

Your First Session

Once Claude Code is installed, here are the first 8 things to try. These cover 90% of daily use.

1

Start Claude Code in any project

Navigate to any project folder and type claude. You'll see a welcome screen with recent conversations and updates.

$ cd ~/projects/my-app && claude

Welcome to Claude Code! Type /help for commands.

>

2

Ask what the project does

Claude reads your files as needed — you don't need to paste anything manually.

> what does this project do?

Reading 32 files...

This is a Node.js REST API with Express...

> what technologies does this project use?

> where is the main entry point?

> explain the folder structure

3

Make your first code change

Claude always shows proposed changes and asks for approval before editing any files.

> add a hello world function to the main file

Found: src/index.js

Proposed change: Add helloWorld() function

Accept? [y/n] y

✓ File edited.

4

Use Git with Claude

Claude Code works conversationally with git — no need to remember commands.

> what files have I changed?

> commit my changes with a descriptive message

> create a new branch called feature/user-auth

> show me the last 5 commits

> help me resolve merge conflicts

5

Fix bugs with error messages

Paste any error or describe the symptom. Claude traces through your code to find the root cause.

> there's a bug where users can submit empty forms - fix it

> add input validation to the user registration form

6

Write tests

> write unit tests for the calculator functions

7

Refactor code

> refactor the authentication module to use async/await instead of callbacks

8

Key keyboard shortcuts

? — show shortcuts

Tab — command completion

↑ — command history

/ — see slash commands

Shift+Tab — cycle permission modes

Ctrl+C — exit / cancel

// daily practice

Common Workflows

Real step-by-step workflows from the official docs. Each includes the exact prompts to type and what to expect back.

🔍 Understand New Codebases

Just joined a new team? First day on a project? Use this to get up to speed in minutes.

Onboarding workflow

# Step 1: Get a high-level overview

> give me an overview of this codebase

# Step 2: Dive into architecture

> explain the main architecture patterns used here

> what are the key data models?

> how is authentication handled?

# Step 3: Find specific code

> find the files that handle user authentication

> how do these authentication files work together?

> trace the login process from front-end to database

# Step 4: Reference files with @ symbol

> Explain the logic in @src/utils/auth.js

> What's the structure of @src/components?

💡

Pro Tip

Start with broad questions, then narrow down. Ask for a glossary of project-specific terms. Use domain language from the project for better results.

🐛 Fix Bugs Efficiently

Paste the error. Let Claude trace it. Ask "why" not just "fix it" — that's how you grow.

Debugging workflow

# Step 1: Share the full error

> I'm seeing an error when I run npm test

TypeError: Cannot read property 'map' of undefined

at ProductList (ProductList.jsx:12)

# Step 2: Get fix suggestions

> suggest a few ways to fix the @ts-ignore in user.ts

# Step 3: Apply the fix

> update user.ts to add the null check you suggested

# Best practice: always ask WHY

> fix the bug and explain what caused it so I can understand it

🎯

Debugging Tips

Tell Claude the exact command to reproduce the issue. Mention if the error is intermittent or consistent. Include the full stack trace — not just the error message.

🧪 Write Tests

From "find uncovered code" to "run and fix" — all in one conversation.

Testing workflow

# Find what's missing

> find functions in NotificationsService.swift that are not covered by tests

# Generate test scaffolding

> add tests for the notification service

# Add edge cases

> add test cases for edge conditions in the notification service

# Run and fix

> run the new tests and fix any failures

✓ 18 tests added. 18/18 passing.

♻️ Refactor Code

Refactoring workflow

# Find legacy code

> find deprecated API usage in our codebase

# Get suggestions

> suggest how to refactor utils.js to use modern JavaScript features

# Apply safely

> refactor utils.js to use ES2024 features while maintaining the same behavior

# Always verify after refactoring

> run tests for the refactored code

⚠️

Safe Refactoring

Always request that changes maintain backward compatibility when needed. Do refactoring in small, testable increments. Ask Claude to explain the benefits of the modern approach.

🔀 Git & Pull Requests

PR workflow

# Create a PR

> summarize the changes I've made to the authentication module

> create a pr

# Enhance the PR

> enhance the PR description with more context about the security improvements

> add information about how these changes were tested

✓ PR #47 created: "feat: Add OAuth2 authentication flow"

🖼️ Work With Images

Drag in a screenshot, mockup, or diagram and Claude can analyze it and generate code from it.

Image workflow

# 3 ways to add images:

1. Drag and drop into the Claude Code window

2. Copy image → Ctrl+V (NOT Cmd+V)

3. Type a path: "Analyze this: /path/to/image.png"

# Then ask about it:

> Here's a screenshot of the error. What's causing it?

> Generate CSS to match this design mockup

> What HTML structure would recreate this component?

> This is our current database schema. How should we modify it for the new feature?

// persistent context

Memory & CLAUDE.md

Claude Code can remember your preferences across sessions — coding styles, common commands, team rules. You control what it remembers and where it's stored.

4 Memory Types — Understand the Hierarchy

Type	File Location	Shared With	Use For
Enterprise Policy	/Library/Application Support/ClaudeCode/CLAUDE.md (macOS) /etc/claude-code/CLAUDE.md (Linux)	All users in org	Company-wide coding standards, security policies
Project Memory	./CLAUDE.md or ./.claude/CLAUDE.md	Team (via git)	Project architecture, team conventions, workflows
User Memory	~/.claude/CLAUDE.md	Just you (all projects)	Personal style preferences, shortcuts
Local Project	./CLAUDE.local.md	Just you (this project)	Your sandbox URLs, test data, local paths

🔒

Auto .gitignore

CLAUDE.local.md is automatically added to .gitignore — perfect for personal preferences that shouldn't be committed to the repo.

Working with Memory

Memory commands

# Fastest way: start input with #

> # Always use 2-space indentation

Select memory file: [1] User [2] Project [3] Local

# Edit memory files directly

> /memory

Opens CLAUDE.md in your system editor

# Bootstrap CLAUDE.md for a new project

> /init

✓ Created CLAUDE.md with project structure info

# See what memory files are loaded

> /memory

Example CLAUDE.md File

# Project: My E-Commerce API

## Build & Test Commands
- Run tests: `npm test`
- Build: `npm run build`
- Lint: `npm run lint`

## Code Style
- Use 2-space indentation
- Prefer async/await over callbacks
- Always add JSDoc for public functions
- Use descriptive variable names — no abbreviations

## Architecture
- Express REST API with PostgreSQL
- Auth via JWT tokens (see src/auth/)
- All DB queries use the ORM in src/db/

## Import other docs
See @README for full overview and @package.json for scripts
Git workflow: @docs/git-instructions.md
  

// built-in commands

Slash Commands

Type / in any session to see all available commands. Here are the most useful ones.

Command	What It Does	When To Use
/help	Show all commands and keyboard shortcuts	First thing to try anytime
/clear	Clear conversation history (free up context)	Starting a new unrelated task
/init	Bootstrap a CLAUDE.md for this project	First time using in a project
/memory	Open memory files in your editor	Update preferences and project rules
/resume	Pick a previous conversation to continue	Coming back to unfinished work
/login	Log in or switch accounts	Switching between work/personal accounts
/mcp	Show and manage MCP server connections	Connecting to GitHub, Sentry, databases
/agents	View, create, and manage subagents	Setting up specialized AI assistants

Create Your Own Slash Commands

Create repeatable workflows as Markdown files. They become slash commands automatically.

Custom commands

# Create project command directory

$ mkdir -p .claude/commands

# Create a "security-review" command

$ echo "Review this code for security vulnerabilities including: SQL injection, XSS, exposed secrets, missing auth checks" > .claude/commands/security-review.md

# Now use it in Claude Code

> /security-review

# Commands with arguments — use $ARGUMENTS placeholder

$ echo "Find and fix issue #$ARGUMENTS: understand it, locate relevant code, implement fix, add tests, write PR description" > .claude/commands/fix-issue.md

> /fix-issue 123

# Personal commands (all projects) — store in home dir

$ mkdir -p ~/.claude/commands

// safe exploration

Plan Mode

Plan Mode lets Claude analyze your codebase with read-only operations — no files get changed. Perfect before making big changes or learning a new codebase.

🔍

Code Exploration

Research the codebase thoroughly without changing anything. Safe to run anytime.

📋

Multi-Step Planning

Plan changes that touch many files before any edits happen. Review the plan first.

🔄

Iterate on Direction

Discuss and refine the approach with Claude before it starts writing code.

Plan Mode usage

# Start a new session directly in Plan Mode

$ claude --permission-mode plan

# Switch modes during a session with Shift+Tab

Normal Mode → Auto-Accept Mode → Plan Mode

Look for: ⏸ plan mode on at the bottom

# Run a one-off analysis in Plan Mode (no edits)

$ claude --permission-mode plan -p "Analyze the authentication system and suggest improvements"

# Example: Plan a complex refactor

> I need to refactor our authentication system to use OAuth2. Create a detailed migration plan.

Reading auth files...

Plan: 1. Update dependencies 2. Create OAuth2 provider 3. ...

> What about backward compatibility?

> How should we handle database migration?

⚙️

Set Plan Mode as Default

Add to .claude/settings.json: {"permissions": {"defaultMode": "plan"}} — useful for teams that want to review all changes before they happen.

// tool integrations

MCP — Connect External Tools

Model Context Protocol (MCP) is an open standard that lets Claude Code connect to hundreds of external tools — GitHub, Sentry, databases, Slack, Figma, Jira, and more.

What You Can Do With MCP

🎫

Issue Trackers

"Add the feature described in JIRA issue ENG-4521 and create a PR on GitHub."

🔥

Monitoring

"Check Sentry for the most common errors in the last 24 hours."

🗄️

Databases

"Find emails of users who used feature X based on our Postgres database."

🎨

Design Tools

"Update our email template based on the new Figma designs posted in Slack."

Installing MCP Servers

HTTP is recommended for cloud-based services. Most popular MCP servers use this transport.

# Basic syntax

$ claude mcp add --transport http <name> <url>

# Real examples

$ claude mcp add --transport http github https://api.githubcopilot.com/mcp/

$ claude mcp add --transport http sentry https://mcp.sentry.dev/mcp

$ claude mcp add --transport http notion https://mcp.notion.com/mcp

# With auth header

$ claude mcp add --transport http myapi https://api.example.com/mcp \

--header "Authorization: Bearer YOUR_TOKEN"

stdio servers run locally on your machine — great for tools needing direct system access.

# The "--" separates Claude's flags from the server's command

$ claude mcp add --transport stdio airtable \

--env AIRTABLE_API_KEY=YOUR_KEY -- npx -y airtable-mcp-server

# Connect to Postgres database

$ claude mcp add --transport stdio db -- npx -y @bytebase/dbhub \

--dsn "postgresql://user:pass@localhost:5432/mydb"

# Windows: Always use cmd /c wrapper

PS> claude mcp add --transport stdio my-server -- cmd /c npx -y @some/package

# List all configured servers

$ claude mcp list

# Get details for a server

$ claude mcp get github

# Remove a server

$ claude mcp remove github

# Check status inside Claude Code

> /mcp

# 3 scope levels:

local (default) — just you, current project

project — shared via .mcp.json in git

user — you, across all projects

$ claude mcp add --scope project --transport http myapi https://...

# ── GitHub: Code reviews & issues ──

$ claude mcp add --transport http github https://api.githubcopilot.com/mcp/

> /mcp (authenticate)

> Review PR #456 and suggest improvements

> Create a new issue for the bug we just found

# ── Sentry: Debug production errors ──

$ claude mcp add --transport http sentry https://mcp.sentry.dev/mcp

> What are the most common errors in the last 24 hours?

> Which deployment introduced these new errors?

# ── Postgres: Query your database ──

$ claude mcp add --transport stdio db -- npx -y @bytebase/dbhub --dsn "postgresql://..."

> What's our total revenue this month?

> Find customers who haven't purchased in 90 days

🛡️

Security Warning

Use third-party MCP servers at your own risk — Anthropic has not verified all of them. Be especially careful with servers that could fetch untrusted content (prompt injection risk). Only install MCP servers you trust.

// specialized helpers

Subagents — AI Specialists

Subagents are specialized AI personalities Claude Code can delegate tasks to. Each has its own context window, tools, and expertise — like a team of specialists.

Built-In Subagents

🔍

Explore Subagent

haiku — read-only — fast

Used automatically when Claude needs to search your codebase without making changes. Strictly read-only — uses Glob, Grep, Read, and limited Bash commands. Multiple thoroughness levels: Quick → Medium → Very Thorough.

GlobGrepReadBash (read-only)

🛠️

General-Purpose Subagent

sonnet — full access

For complex multi-step tasks requiring both exploration and file modification. Has access to all tools. Used when tasks involve multiple steps that depend on each other.

All toolsRead + WriteCommands

📋

Plan Subagent

sonnet — plan mode only

Used automatically in Plan Mode to research your codebase before presenting a plan. Can explore files but cannot execute changes.

ReadGlobGrepBash

Create Your Own Subagents

Creating custom subagents

# Interactive way (recommended)

> /agents

→ Select "Create New Agent" → Generate with Claude first

# Manual: Create a code-reviewer subagent

$ mkdir -p .claude/agents

$ cat > .claude/agents/code-reviewer.md <<'EOF'

---

name: code-reviewer

description: Expert code review specialist. Proactively reviews code for quality, security, and maintainability. Use immediately after writing or modifying code.

tools: Read, Grep, Glob, Bash

model: inherit

---

You are a senior code reviewer...

EOF

# Invoke explicitly

> Use the code-reviewer subagent to check my recent changes

> Have the debugger subagent investigate this error

✨

Pro Tip: Make Delegation Proactive

Add "Use PROACTIVELY" or "MUST BE USED" in your agent's description field to encourage Claude to delegate automatically without being asked explicitly.

// automation

Hooks — Auto-Trigger Actions

Hooks let you run shell commands automatically before or after Claude Code performs certain actions. Think of them as event listeners for your coding workflow.

PreToolUse

Run before Claude uses a tool. Can block the action. Example: auto-run prettier before any file edit.

PostToolUse

Run after Claude uses a tool. Example: auto-lint after every file save, or notify Slack on commit.

Notification

Triggered when Claude needs user attention. Example: play a sound when Claude finishes a long task.

Stop

Run when Claude finishes responding. Example: run your test suite automatically after each response.

Example Hooks Config

// .claude/settings.json
{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "npm run lint -- --fix"
          }
        ]
      }
    ],
    "Stop": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "npm test"
          }
        ]
      }
    ]
  }
}
  

// unix philosophy

Headless Mode & Scripting

Claude Code is composable and scriptable. Pipe data in, pipe results out. Integrate into CI/CD. Chain with other Unix tools.

CLI scripting examples

# One-off task (non-interactive)

$ claude "fix the build error"

# Run query then exit (-p flag)

$ claude -p "explain this function"

# Continue most recent conversation

$ claude --continue

$ claude --continue --print "Continue with my task"

# Pipe data in, get results out

$ cat build-error.txt | claude -p 'explain the root cause of this build error' > output.txt

# Monitor logs in real time

$ tail -f app.log | claude -p "Slack me if you see any anomalies"

# Automate in CI (translate strings, open PR)

$ claude -p "translate new strings into French and raise a PR for @lang-fr-team to review"

# Bulk security review on changed files

$ git diff main --name-only | claude -p "review these changed files for security issues"

# JSON output for scripts

$ cat code.py | claude -p 'analyze for bugs' --output-format json > analysis.json

📄

--output-format text

Default. Returns just Claude's plain text response. Good for simple integrations.

📊

--output-format json

Full conversation log as JSON array with metadata including cost and duration.

📡

--output-format stream-json

Real-time stream of JSON objects. Each message is valid JSON as Claude processes.

// ci/cd automation

GitHub Actions Integration

Tag @claude in any PR or issue comment — Claude will analyze, implement, fix bugs, and open PRs automatically. All without leaving GitHub.

What Claude Can Do in GitHub

✅ Create complete PRs

🐛 Fix reported bugs

🔍 Review code changes

📝 Respond to comments

🧪 Write tests

📖 Update documentation

Quick Setup

Automated setup

# Easiest: use the built-in installer

$ claude

> /install-github-app

→ Follow prompts to install GitHub app and set secrets

# Requires: repo admin access

App needs: Contents R/W, Issues R/W, Pull Requests R/W

Basic Workflow File

# .github/workflows/claude.yml
name: Claude Code
on:
  issue_comment:
    types: [created]
  pull_request_review_comment:
    types: [created]
jobs:
  claude:
    runs-on: ubuntu-latest
    steps:
      - uses: anthropics/claude-code-action@v1
        with:
          anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
          # Claude responds to @claude mentions automatically
  

How to Use in Comments

GitHub PR/Issue comments

# In any PR or issue comment:

@claude implement this feature based on the issue description

@claude how should I implement user authentication for this endpoint?

@claude fix the TypeError in the user dashboard component

@claude review this PR and suggest improvements

# Claude will:

1. Read the issue/PR context

2. Analyze your codebase

3. Create a branch, implement changes, open a PR

4. Comment back with what it did

⚡ Git Worktrees — Parallel Sessions

Need to work on a bugfix AND a feature at the same time? Git worktrees give each Claude session its own isolated code copy.

Parallel Claude sessions

# Create isolated worktrees

$ git worktree add ../project-feature-a -b feature-a

$ git worktree add ../project-bugfix bugfix-123

# Run Claude in each (separate terminal windows)

$ cd ../project-feature-a && claude

$ cd ../project-bugfix && claude

# Both Claude sessions work independently!

project-feature-a: Claude building new user auth...

project-bugfix: Claude fixing cart calculation bug...

# Manage worktrees

$ git worktree list

$ git worktree remove ../project-feature-a

// quick reference

Full Cheatsheet

Every command, flag, and prompt you'll need. Save this page or bookmark it.

CLI Commands

Command	What It Does	Example
claude	Start interactive mode	cd my-project && claude
claude "task"	Run one-time task	claude "fix the build error"
claude -p "query"	One-off query, then exit	claude -p "explain this function"
claude -c	Continue most recent conversation	claude -c
claude -r	Resume a conversation (picker)	claude -r
claude commit	Create a git commit	claude commit
claude --permission-mode plan	Start in Plan Mode (read-only)	claude --permission-mode plan
claude mcp add	Add an MCP server	claude mcp add --transport http github https://...
claude mcp list	List connected MCP servers	claude mcp list

High-Value Prompts by Category

give me an overview of this codebase

Onboarding magic. Works on any project. Gets you up to speed in minutes.

understand

find the files that handle [feature]

Code discovery. "find the files that handle user authentication" — then drill down.

explore

trace [process] from front-end to database

End-to-end understanding. Trace the login process. Trace the checkout flow.

explore

explain the logic in @src/utils/auth.js

File-specific explanation. The @ symbol inlines the file content automatically.

@reference

fix the bug and explain what caused it

Learn while fixing. Don't just fix — understand why it broke. This is how you improve.

debug

suggest a few ways to fix [issue]

Options before committing. See multiple approaches before Claude edits anything.

debug

add [feature] following the existing code style

Style-consistent features. Claude reads your patterns and matches them exactly.

feature

find functions without tests and write them

Instant coverage. Claude finds uncovered code AND writes the tests in one go.

testing

run the tests and fix any failures

Self-healing tests. Claude runs, reads the errors, and fixes them autonomously.

testing

review my git diff for bugs and security issues

Pre-push safety check. Catches what you'd miss. Run before every commit.

security

commit my changes with a good message

Perfect git commits. Stages files and writes conventional commit messages automatically.

git

create a pr with a full description

Auto PR writing. Branch, commit, push, and open a complete PR with context.

git

refactor [file] — don't change behavior

Clean up safely. Improves readability while preserving the same functionality.

refactor

think hard about [complex problem]

Extended thinking. "think" = basic, "think hard" = deep, "think a lot" = maximum depth.

thinking

# always use 2-space indentation

Quick memory save. Start with # to instantly save a preference to CLAUDE.md.

memory

// learn from others

Common Mistakes

These are the exact mistakes every junior dev makes in their first weeks with Claude Code. Learn them now.

❌ Don't Do This

fix my code

Too vague. Claude doesn't know what's broken, where, or what "fixed" looks like.

✅ Do This Instead

fix the 401 error in auth.js line 47 — users get logged out when their token refreshes

Specific file, line, symptom, and expected behavior. Claude nails it first try.

❌ Don't Do This

build my entire app for me

Too broad. Claude does better with focused, clear scope per request.

✅ Do This Instead

build the user registration form. Match the LoginForm.jsx pattern we already have.

Small, focused scope + reference existing patterns = high-quality output.

❌ Don't Do This

Accept all changes without reading them

You're responsible for code in your repo. Blindly accepting leads to issues at review time.

✅ Do This Instead

Ask "explain what you changed and why" before accepting

Understanding changes makes you a better developer AND protects you at code review.

❌ Don't Do This

Never run the code after Claude edits

Claude is good but not perfect. Always verify changes work in your actual environment.

✅ Do This Instead

run the tests and verify the app still works after your changes

Ask Claude to test too — it can run your test suite and fix failures automatically.

❌ Don't Do This

Start new chat for every single question

Each new session loses context. Claude doesn't know your project anymore.

✅ Do This Instead

Keep the session going. Use /clear only when switching topics.

A long session where Claude knows your whole project is 10x more powerful.

// your journey

30-Day Roadmap

Follow this plan and in one month you'll go from "just installed it" to power user level.

W1

Week 1 — Install, Explore, Understand

Get Claude Code running. Spend the whole week just asking it to explain codebases — yours, open source projects, tutorials. Use /init on one project. Get comfortable with the conversation flow. Don't worry about making changes yet.

installcodebase exploration/initasking questions

W2

Week 2 — Debug & Fix Bugs

Find a bug in something you're working on — or create one intentionally — and use Claude to trace and fix it. Always ask "explain what caused it" to learn. Try pasting error messages. Try the Plan Mode exploration approach first.

debuggingplan modeunderstanding root cause

W3

Week 3 — Build Features & PRs

Use Claude to implement a real feature from your backlog. Practice the plan-then-build workflow. Always run Claude's code review before pushing. Get your first Claude-assisted PR merged.

feature devcode reviewgit workflowPRs

W4

Week 4 — Tests, Memory & Automation

Write tests for a module that has poor coverage. Set up a CLAUDE.md for your project. Try connecting one MCP server (GitHub or Sentry). Set up a GitHub Action. Create one custom subagent.

testingCLAUDE.mdMCPGitHub Actionssubagents

🎉

Day 30 — Power User Status

You now use Claude Code without thinking about it. You ship features faster, write better code, have better test coverage, and understand new codebases in minutes. You're using memory, agents, and MCP. You look like a senior dev.

🚀 Level up complete!

Ready to start right now?

Open your terminal. Navigate to any project. Type claude. That's the whole step.

Sign up at claude.ai → Official Docs ↗

All information sourced directly from code.claude.com/docs

MasterClaude Code

Installation & Setup

Open Extensions Panel

Search "Claude Code"

Open Claude Code

Your First Session

Start Claude Code in any project

Ask what the project does

Make your first code change

Use Git with Claude

Fix bugs with error messages

Write tests

Refactor code

Key keyboard shortcuts

Common Workflows

🔍 Understand New Codebases

🐛 Fix Bugs Efficiently

🧪 Write Tests

♻️ Refactor Code

🔀 Git & Pull Requests

🖼️ Work With Images

Memory & CLAUDE.md

4 Memory Types — Understand the Hierarchy

Working with Memory

Example CLAUDE.md File

Slash Commands

Create Your Own Slash Commands

Plan Mode

MCP — Connect External Tools

What You Can Do With MCP

Installing MCP Servers

Subagents — AI Specialists

Built-In Subagents

Create Your Own Subagents

Hooks — Auto-Trigger Actions

Example Hooks Config

Headless Mode & Scripting

GitHub Actions Integration

What Claude Can Do in GitHub

Quick Setup

Basic Workflow File

How to Use in Comments

⚡ Git Worktrees — Parallel Sessions

Full Cheatsheet

CLI Commands

High-Value Prompts by Category

Common Mistakes

30-Day Roadmap

Week 1 — Install, Explore, Understand

Week 2 — Debug & Fix Bugs

Week 3 — Build Features & PRs

Week 4 — Tests, Memory & Automation

Day 30 — Power User Status

Ready to start right now?

Frequently Asked Questions

Master
Claude Code