Published notes
When CodeRabbit Catches What Agents MissA historical code-review lesson, using an illustrative 17.6-to-18.4 upward boundary jump to show why piecewise scoring functions need explicit seam tests.7 min read When Your Calibrator Is Silently ZeroA historical scoring-system lesson: default outputs can conceal a weak or inactive component, so contribution-rate monitoring should complement tests and logs.8 min read