Dataset Cleaning and Quality Agent

Most analysis time goes to cleaning messy data by hand, and errors slip into results. Before any analysis this agent profiles a new dataset against your rules and lists each problem with counts and examples: missing values, duplicates, wrong types and impossible numbers. It applies safe fixes on its own, such as trimming spaces or standardizing dates. For risky changes, like dropping rows or filling missing values, it proposes the change, shows how many records it affects and waits. After cleaning it reruns every check to confirm the problems are gone and no new ones appeared. It writes a cleaning log so the work can be repeated next month. You approve the risky fixes. Edge case: a real outlier, like a genuine large order, is kept and noted, not removed.