Trace To Training Data This skill assumes already graded the traces being converted here — goldens, graders, and all exist before conversion starts. This is the flywheel edge that skill names in its own flow: "the same labeled traces become the training set." Conversion happens here; grading already happened upstream. Input: graded traces — plus , each row carrying a , a from the grader, and a when the task supports a scalar score (judge score, execution partial-credit, or an RLVR verifier): Output format: rows shaped exactly like 's Format Selection table — SFT rows or DPO / / pairs — so thi…