Prompt Engineering gets its own section

Prompt Engineering is now a first-class section in /learn — three pillars, twelve guides, covering the full range from first principles through advanced techniques to building with prompts in production.

The new techniques pillar ships four guides: chain-of-thought (when showing your reasoning changes the output), prompt chaining (how to split complex tasks into reliable steps), role prompting (when persona assignment actually steers behavior vs when it's theater), and structured output control (getting JSON, tables, and formats that hold).

The foundations pillar carries the four existing guides on prompt basics, XML structure, and few-shot examples — moved here where they belong.

System prompts, iteration method, injection defense, and cross-model porting are queued for the for-builders pillar.