The Unreasonable Effectiveness of Category Theory in Preventing LLM Drift

Can Category Theory tame the “AI slop” that coding agents produce? We pressure-test the thesis that a CT-derived skeleton lets an LLM iterate toward production code while catching intent drift at checkpoints.