C1: Context Engineering Competency
Shopify CEO Tobi Lutke noted: "I really like the term 'context engineering' over prompt engineering. It describes the core skill better: the art of providing all the context for the task to be plausibly solvable by the LLM." Andrej Karpathy compares the LLM context window to RAM, a limited working memory that requires careful curation of what fits.
Transformer architecture computes relationships between every token pair, creating n-squared complexity. As context grows, models experience "context rot" where accuracy degrades because attention spreads thin across more tokens. The effective context window where models perform at high quality is often much smaller than the advertised token limit.
Principle
- All roles MUST develop Context Engineering competencies, maintaining clean context for AI consumption and avoiding noise and irrelevant information.
Only include information pertinent to the task at hand. Context engineering means choosing what NOT to include as much as what to include. Organize context with clear hierarchy and consistent formatting. A codebase should not contain multiple patterns for the same problem. Remove outdated information and ensure synchronization with code. Inconsistent conventions create noise that degrades AI output quality.