Claude Code Workflow Cheatsheet — Orbytt Academy
Orbytt Academy · Claude Code

Workflow Cheatsheet for Beginners

Everything you need to start using Claude Code with confidence — explained in plain English, no technical background required.

Getting Started · Project Setup · File Structure · Skills · Hooks · Memory · Workflows · 2026 Edition
1
Getting Started

Install Claude Code — requires Node.js 18+

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

Navigate to your project folder and launch:

cd your-project
claude
/init

/init scans your codebase and creates a starter memory file automatically.

2
Understanding CLAUDE.md

CLAUDE.md = Claude’s persistent memory about your project. Loaded automatically at the start of every session.

What
  • Tech stack
  • Directory map
  • Architecture
Why
  • Module purpose
  • Design choices
  • Gotchas
How
  • Build/test/lint
  • Workflows
  • Commands
# Project: MyApp
FastAPI REST + React SPA + Postgres

## Commands
npm run dev / test / lint

## Architecture
/app -> Next.js pages
/lib -> shared utilities
/prisma -> DB schema
3
Memory File Hierarchy

Multiple CLAUDE.md files stack at different levels — deeper files add context without overwriting the parent.

~/.claude/CLAUDE.md Global — rules that apply to all your projects ~/CLAUDE.md Parent — monorepo root, shared across sub-projects ./CLAUDE.md Project — committed to Git, shared with your team ./frontend/CLAUDE.md Subfolder — scoped context for one part of the app · Keep each file under 200 lines    · Subfolder files append context    · Never overwrite parent
4
CLAUDE.md Best Practices
  • Run /init first, then refine the output
  • Be specific — vague prompts get vague results
  • Add gotchas Claude cannot infer from the code alone
  • Reference specific files with @filename
  • Add workflow rules so Claude follows your team’s process
  • Keep memory concise — remove outdated notes regularly
  • Commit CLAUDE.md to Git so your whole team benefits
5
Project File Structure
your-project/ CLAUDE.md .claude/ settings.json settings.local.json skills/ code-review/ SKILL.md testing/ SKILL.md commands/ deploy.md agents/ Folder Memory file Skill / config files
6
Adding Skills — The Superpower

Skills = markdown guides Claude auto-invokes via natural language. The description field is critical for auto-activation.

Project skill .claude/skills/name/SKILL.md Shared with team via Git Personal skill ~/.claude/skills/name/SKILL.md Available across all your projects name: testing patterns description: Jest testing patterns | allowed tools: Read, Grep, Glob # Use describe+it+AAA pattern. Use factory mocks.
7
Skill Ideas for AI Engineers

Start with these — each becomes a markdown file Claude learns to invoke automatically when you describe your intent.

code-review
docker-deploy
testing patterns
codebase-visualizer
commit messages
api-design
8
Setting Up Hooks

Hooks = deterministic callbacks that run automatically before or after Claude uses a tool.

PreToolUse PostToolUse Notification
"hooks": {
  "PreToolUse": [{
    "matcher": "Bash",
    "hooks": [{
      "type": "command",
      "command": "scripts/sec.sh",
      "timeout": 5
    }]
  }]
}
Exit codes  →  0 = allow  ·  2 = block
9
Permissions & Safety

Control exactly what Claude is and isn’t allowed to do using a permissions block in your settings file.

{
  "permissions": {
    "allow": [
      "Read:*",
      "Bash:git:*",
      "Write:*:*.md"
    ],
    "deny": [
      "Read:env:*",
      "Bash:sudo:*"
    ]
  }
}

Allows: reading any file, using git, writing Markdown files.
Blocks: reading environment variables, running admin commands.

10
The 4-Layer Architecture
L1 CLAUDE.md Persistent context and rules — always loaded first L2 Skills Auto-invoked knowledge packs L3 Hooks Safety gates and automation triggers L4 Agents Subagents with their own independent context
11
Daily Workflow Pattern
1 Open your project and start Claude cd project && claude 2 Switch to Plan Mode so Claude thinks before acting Shift+Tab+Tab 3 Describe the feature or fix you want to build type your intent 4 Enable Auto Accept to skip manual confirmations Shift+Tab 5 Compress the context to free up working space /compact 6 Undo if something went wrong Esc Esc → rewind 7 Commit your progress frequently git commit 8 Start a fresh session for each new feature new session
12
Quick Reference
Command / Key What It Does /init Generate CLAUDE.md by scanning your project /doccat Check that Claude Code is installed correctly /compact Compress the context window to save space Shift + Tab Cycle modes: Normal → Plan → Auto Accept Tab Toggle extended thinking on / off Esc Esc Open rewind menu to undo recent changes Tip: Shift+Tab+Tab jumps straight into Plan Mode without cycling