From prompt engineering to context engineering — budgets, retrieval, compression, and company-wide repository context.
The industry discussion has shifted from prompt engineering toward context engineering: selecting, structuring, and budgeting everything an agent sees each step — policy, AGENTS.md, retrieved code, tool results, and memory.
An important nuance from current research: concise, focused context (including AGENTS.md) helps, but overly long or overly prescriptive context files can increase inference cost and reduce task success. That makes 'best practices' content more valuable than generic tutorials — and it is exactly what interviews probe.
This guide defines context engineering, contrasts it with prompt engineering, and walks through repository context systems at company scale. Pair with the AGENTS.md guide for implementation detail.
Deep explanations with architecture diagrams for every question below.
Prompt engineering focuses on instruction wording. **Context engineering** focuses on the full state fed to the model each step: policies, repository maps, retrieved files, prior tool outputs, memories, and task specs.
Read full explanationCompany-wide context is a platform product.
Read full explanation