promptdojo_

Four prompts written today. Three of them are clean — they transfer information the model can't infer (stack, schema, examples, format). One of them is leaning on a 2023-era technique that's now noise on a modern reasoning-capable frontier model.

Which one?

read, then continue.