Foreword
Software rarely comes alone these days. A single container image contains an operating system, a runtime, dozens of libraries and your team’s code. Each of these parts has an origin, and each of them can be altered, poisoned or simply outdated on its way to you. If you are responsible for such a supply chain, you need to be able to do more than “make the pipeline green”.
This book is not a reference to leaf through. It is a drill course. You read little, train a lot and go home with muscle memory. At the end you can not only explain the ten skills, you can demonstrate them. These exact ten show up again and again as must-haves in job postings for security-critical platform and DevOps roles:
| Discipline | What you train | Chapter |
|---|---|---|
| 1 | Running CI/CD chains yourself (GitLab CI, Jenkins, Argo CD/Flux) | 1 |
| 2 | Assessing and hardening existing build processes | 2 |
| 3 | Mirrored sources as the only supply path (Harbor, Nexus, Artifactory) | 3 |
| 4 | Controlled artifact transfer across zone boundaries | 4 |
| 5 | Offline signing and provenance verification with your own trust anchor | 5 |
| 6 | Secrets management incl. rotation (Vault/OpenBao, External Secrets) | 6 |
| 7 | SBOM in CycloneDX/SPDX and dependency analysis | 7 |
| 8 | SAST, SCA, image and IaC scans incl. triage and thresholds | 8 |
| 9 | Policy as code (Kyverno, OPA) | 9 |
| 10 | Working with development teams, exceptions, transition periods | 10 |
The training rules
- If you don’t type, you don’t train. Type the commands, don’t copy them. Then change a parameter and watch what happens.
- Errors are training stimuli. When something turns red, that is not a setback, it is the actual session. Read the pitfalls table first, then search on your own, and only after that look at the solution.
- No sparring without a warm-up. Read the analogy at the start of the chapter before you go into the lab. If you have the picture in your head, you find errors twice as fast.
- Explain out loud. After each session you answer the interview questions out loud, as if someone were sitting across from you. What you cannot explain in two sentences, you have not mastered yet.
- Lab values stay in the lab. All passwords, tokens and keys are training equipment. Never use them in production.
The flow of each training session
Each chapter is a training session with a fixed flow:
| Phase | Section in the chapter | Goal |
|---|---|---|
| Warm-up | What is it about? | The picture in your head: analogy and sketch |
| Technique | The concepts made simple | Only the terms you really need |
| Drill | Hands-on Lab | Step by step, with expected output |
| Error analysis | Common pitfalls | Recognize the symptom, find the cause |
| Sparring | Exercise | Solve it alone, only then hint 1, hint 2, solution |
| Competition practice | From the field | How it works in real, regulated environments |
| Exam | Interview questions | Answer in your own words |
| Cool-down | Summary | What should stick |
The training plan
The course is designed for three weeks with five training days each, at about two to three hours per day. If you are faster: good. If a session takes longer: also good. Repeat it rather than rushing on.
| Week | Day | Session | Focus |
|---|---|---|---|
| 1 – Foundations | 1 | Chapter 0 | Build and check the lab |
| 2–3 | Chapter 1 | Running CI/CD yourself | |
| 4 | Chapter 2 | Assessing build processes | |
| 5 | Chapter 3 | Mirror as the only supply path | |
| 2 – Supply chain | 6 | Chapter 4 | Transfer across zone boundaries |
| 7–8 | Chapter 5 | Offline signing and your own PKI | |
| 9 | Chapter 6 | Secrets and rotation | |
| 10 | Chapter 7 | SBOM and dependencies | |
| 3 – Enforcement | 11 | Chapter 8 | Scans, triage, thresholds |
| 12 | Chapter 9 | Policy as code | |
| 13 | Chapter 10 | Exceptions and transition periods | |
| 14 | Chapter 11 | Competition day: the whole chain | |
| 15 | Appendix C | Final exam and self-assessment |
Tip: Put the training days in your calendar. A fixed time slot beats every good intention.
The training ground
All sessions run on your own machine: Docker, a small Kubernetes
cluster with kind, two local registries that represent two
security zones, and a tiny sample application demo-app that
accompanies you through the whole course. Chapter 0 builds the training
ground. From Chapter 1 on, you can train the sessions largely
independently of each other. Chapter 11 is competition day: there you
connect everything into one continuous, secured chain.
Internet Zone "Build/Internal" Zone "Target"
┌───────────┐ Mirror ┌──────────────────────────┐ Airlock ┌──────────────────┐
│ Docker Hub│──────────► │ Nexus ─► CI ─► reg-build │ ───────► │ reg-ziel ─► kind │
│ PyPI, apt │ (Ch. 3) │ Scan · SBOM · Signature │ (Ch. 4) │ Kyverno · Argo CD│
└───────────┘ └──────────────────────────┘ └──────────────────┘
Conventions
- Commands appear in gray boxes.
~/seclabis your training ground. - Remember boxes summarize the core. Warning alerts you to traps. Security marks things you must do differently in real operations than in the lab.
- Tools evolve quickly. If a flag does not exist (anymore),
<tool> --helpand the official documentation help. This kind of looking things up is a core skill of the profession.
Good luck in the lab, and have fun building and breaking things. You learn from both.
Thomas Zachmann, October 2026