This is a book you read at night, one chapter at a time. Each chapter takes 5โ15 minutes to read, and longer if you stop to try things. Read it slowly, and when a chapter shows a command, don't type it now. Tomorrow, when you sit at your laptop, open the Hacking Lab and do it for real. Reading builds the map; the lab builds the reflexes. You need both.
The guide follows the path we agreed on:
| Part | What you learn | Where you practise |
|---|---|---|
| Foundations | Linux, networking, the attacker's method, privilege escalation | OverTheWire Bandit, TryHackMe beginner paths |
| Web | How web apps break, the PortSwigger companion, then every major bug class and its fix | PortSwigger Web Security Academy, Juice Shop, DVWA |
| Defense | How attacks are detected and contained | Your own logs, the Practice Lab breach loop |
| Careers | Security+, OSCP, PNPT, BTL1 | A 12-week plan at the end |
The one rule
Only attack what you own, or what you have written permission to attack.
That's it. Everything in this guide is legal on your own machines, on deliberately vulnerable apps running in your home lab, on CTF platforms (picoCTF, HackTheBox, TryHackMe, PortSwigger labs) and in authorized penetration tests with a signed scope.
The same command run against a website you don't own is a crime. In the US that's the Computer Fraud and Abuse Act; most countries have an equivalent. "I was only looking" or "I didn't change anything" is not a defense. Scanning a stranger's server "to learn" is how people get their first letter from a lawyer.
Some practical consequences:
- Your work laptop and work systems are not yours. Pulse belongs to your employer. You may test it only on your local copy, with your own test accounts, and report findings through the normal channel.
- Production data is never a target. Even in your own projects, attack the local or synthetic version. Maeve holds a real child's medical data; you attack the synthetic preview, never the live family.
- Bug bounties have scopes. If you later hunt on HackerOne or Bugcrowd, the program page lists exactly what's allowed. Out of scope means illegal.
How a hacker thinks
Developers ask: "What is this feature supposed to do?" Attackers ask: "What does this feature actually accept, and what happens if I give it something nobody expected?"
Every vulnerability in this book is a gap between those two questions. A few examples:
- A login form is supposed to take a username. It actually takes any string, and if that string reaches a SQL query unescaped,
admin'--logs you in. (Chapter 6.) - A page is supposed to show your invoice at
/invoice/1043. It actually shows whichever invoice number you type. (Chapter 6.) - A feature is supposed to fetch an image from a URL you give it. It actually fetches any URL, including
http://169.254.169.254/inside the cloud, where the server's secret keys live. (Chapter 6.)
So the attacker's habit is: find every input, and for each one ask what it trusts. Inputs are more than form fields. They include URL paths, query strings, cookies, HTTP headers, file names, JSON keys, the order of requests, and even timing.
The attacker's method
Professional testers follow the same loop every time. You'll see it in Bandit, in TryHackMe, in the OSCP exam and in real engagements:
- Reconnaissance. What exists? Domains, IPs, open ports, technologies, people, public code.
- Enumeration. Go deep on what you found: versions, endpoints, users, shares, parameters.
- Vulnerability analysis. Which of those things are weak? Known CVEs, misconfigurations, logic bugs.
- Exploitation. Prove it: get a shell, read the file, become another user.
- Post-exploitation and privilege escalation. From a low user to root or admin, from one machine to the next.
- Reporting. Explain what you did, the impact, and how to fix it. This is the part you're paid for.
Most beginners rush to step 4. Most experts spend 70% of their time in steps 1 and 2. When you're stuck on a box, the answer is almost always: enumerate more.
Notes: your second brain
Start a notes system on day one. Every tester has one. Options: Obsidian (free, Markdown files), CherryTree, Notion, or a plain folder of .md files. For each machine or challenge, write:
# Target: Bandit level 12 (or HTB "Lame", or Juice Shop "Admin login")
## Recon
- what I ran, what came back (paste the important lines)
## What I tried
- tried X โ failed because Y
## What worked
- the exact command / payload
## Lesson
- one sentence: what I'll recognize faster next time
The "what failed" section matters more than "what worked". In six months you'll search your notes for an error message and find the answer you already learned.
A nightly rhythm
- Night (here): read one chapter. Highlight one thing you didn't know.
- Next free hour (lab): do the exercises linked from that chapter; log it in the Hacking Lab's CTF log.
- Weekend: one bigger thing: a full TryHackMe room, a HackTheBox machine, or a mission against your own code.
Consistency beats intensity. Thirty minutes every night for three months will take you further than a weekend binge.
Using AI as a study partner
Claude is good at explaining why an exploit works, reviewing your notes, suggesting what to enumerate next, and writing small tools (parsers, scanners, solvers) for your lab. To get the most out of it:
- Say the context. "This is a picoCTF challenge", "this is DVWA in my home lab", "this is my local copy of my own app". Context gets you complete, direct answers.
- Paste output, not summaries. The exact nmap result or HTTP response is worth ten descriptions.
- Ask for hints before answers. "Give me a nudge, not the solution" keeps the learning yours.
- Ask it to quiz you. "Ask me five questions on chapter 7" at the end of a reading night is excellent revision.
Turn the page. Chapter 1 starts with the tool you'll use more than any other: the Linux shell.