Audit allocation mechanisms for group and intersectional disparities.
npx skills add https://github.com/AndurilCode/craftwork --skill fairness-auditing
Please help me install this Agent Skill. Skill: fairness-auditing Source: https://github.com/AndurilCode/craftwork/tree/main/skills/fairness-auditing Command: npx skills add https://github.com/AndurilCode/craftwork --skill fairness-auditing
It helps you determine whether a system that allocates outcomes across groups is equitable by selecting explicit fairness definitions, measuring disparities, and identifying where bias enters.
Ask the AI to perform a fairness audit for your allocation system by stating the decision type, affected groups, relevant protected attributes, and the fairness definition(s) you want to test.
High-intent search queries and answers about installing and using this skill.
Auditing algorithmic bias involves selecting fairness definitions like demographic parity, measuring intersectional disparities with significance checks, and identifying historical bias or proxy variables that reconstruct protected attributes.
Disparate impact is measured by computing group-level and intersectional outcome rate ratios, then applying statistical significance checks to quantify disparities across protected demographic groups.
Check for proxy variables by auditing training data for representation gaps and features that reconstruct protected attributes, identifying where historical bias enters the allocation mechanism.
Mitigate algorithmic bias by applying pre-, in-, or post-processing interventions, then assess fairness gains versus accuracy trade-offs and document residual-risk recommendations for the allocation system.
Yes, evaluate intersectional fairness by computing disparities across overlapping demographic groups and applying significance checks to ensure equitable outcomes in algorithmic hiring, lending, or triage models.
Define fairness criteria when outcome rates differ by group membership, clarifying value trade-offs between definitions like demographic parity, equal opportunity, and predictive parity before measuring disparities.
hoavdcAudit datasets across six DAMA dimensions and generate a scored quality report.
praburajasekaranVerify production data deployments with SQL queries and rollback procedures.
lebsralMonitor AI model performance, drift, and cost in production.
TurnaboutHeroAnalyze structured datasets with hypothesis tests, regression models, and confidence intervals.
