The Red Line Principle: objective stop signals outperform LLM self-judgment in verifiable tasks

· Dev.to