Claude AI Prompt Engineering for Engineers & Project Managers
How to structure prompts for technical work — context blocks, role framing, examples and controlled outputs that actually work.
Claude and other frontier LLMs respond dramatically better when you feed them structured context. Here is a template that works across engineering, PM and commercial work.
The C.R.I.S.P. structure
- Context — one paragraph on the project, phase and audience.
- Role — "You are a chartered civil engineer reviewing…".
- Input — the raw material (drawing notes, contract clause, minutes).
- Steps — 3–5 explicit steps for the model to follow.
- Pattern — an example of the desired output.
Use tags for long context
Wrap large inputs in tags like <contract>…</contract> — Claude follows them reliably.
Ask for reasoning, then answer
"Think step by step, then give the final answer under Answer:". Quality jumps immediately.
Iterate, don't restart
Refine with follow-ups instead of rewriting from scratch — Claude preserves context brilliantly.
Recommended course
Our Claude AI for Project Managers course walks through 30+ engineering & PM prompt patterns.
AI for Engineers and Project Professionals
Ready to go beyond one article? Explore the complete learning path built around this topic.
