Prompting and Context Engineering for Infrastructure Work
Most bad AI output is not the model's fault — it is a context and requirements problem. Throwing a vague request over the wall and getting back plausible garbage is the engineering equivalent of writing a one-line ticket and being surprised the feature missed the point. These chapters reframe prompting as something engineers already know how to do: gathering requirements, stating constraints, and supplying the context a system needs to produce something usable.
We treat context as a control plane. The quality of what AI gives you depends heavily on what it can see, what it is allowed to assume, and how well your repos, docs, ownership boundaries, and standards are structured. Get that right and the model becomes genuinely useful; get it wrong and no amount of clever phrasing saves you.
What you’ll learn
- How to give AI useful constraints, architecture context, acceptance criteria, risks, and examples instead of vague requests.
- Why prompting is really requirements gathering, and how to apply skills you already have.
- How repo structure, docs, ownership, and standards shape the quality of AI output.
- What "context is the control plane" means in practice for infrastructure and platform work.
Who it’s for
Engineers who want AI to produce output they can actually use — and platform teams deciding how to structure repos, docs, and standards so AI assistance is reliable rather than random.
Chapters in Prompting & Context Engineering
Drawn from the working table of contents of Confidently Wrong. Subject to revision as the manuscript develops.
Prompting Is Requirements Gathering
How to give AI useful constraints, architecture context, acceptance criteria, risks, and examples instead of tossing vague requests over the wall.
Context Is the Control Plane
Why AI output quality depends heavily on what it can see, what it is allowed to assume, and how well your repos, docs, ownership, and standards are structured.
Read the full argument.
These chapters are part of Confidently Wrong — a practical book for DevOps, SRE, platform, and infrastructure engineers on adopting AI safely without giving it unchecked authority over production.
← Back to the bookWant the same lens in podcast form? Browse the Platform Engineering episodes on Ship It Weekly.