mstevens843/agent-eval-foundry

A methodology and toolkit for turning agent-eval ideas into measurable task families: catch-set analysis, axis diversity scoring, mutant checks, and evidence-backed task selection.