If your local-Ollama agent has been getting quietly worse for no obvious reason — same model, same hardware, same prompts — there's a good chance you're hitting an invisible ceiling that produces no error, no warning, and no log line. Just an empty response where an answer used to be. I want to walk through how this manifests, why it's specifically painful for autonomous agent workloads (not chat)
Ages ago when CSS grids came with a repeat() function to simplify defining repetitive columns and rows, I was not alone in wishing for this function to be made generic and work in any context. After seeing Wes Bos on BlueSky wishing for this exact concept, specifically for repeating segments in a shape() definition, I chimed in with my +1's on making repeat() generic across CSS Without delay, @no