risk-rule-design

In risk strategy analysis, rule design often relies on manual trial and error: checking data quality, evaluating fields one by one, and testing many rule combinations. This process is time-consuming and highly dependent on individual experience. risk-rule-design is a DeepSeek Harness plugin that automates the risk rule mining workflow, helping users move from raw sample data to rule insights and reports in a more structured way.

The plugin covers data profiling, single-rule evaluation, parallel rule-set optimization, and HTML report generation. It helps confirm the usable fields, removes low-value or high-risk columns, assesses candidate rules by metrics such as hit rate, bad rate, and Lift, and then searches for stronger combined rule sets. The final report presents both individual variable effects and combination performance in a visual, browser-openable format.

It is best suited for risk strategy analysts, data analysts, and teams that need a faster, repeatable way to explore features and rules. It does not replace business judgment, but it reduces repetitive analysis work and lets users focus more on strategy interpretation, edge cases, and practical rollout.