Function Calling Harness 2: CoT Compliance from 9.91% to 100%

TL;DR 9.91% is not "did the model get it right on the first try" — it's "did the model walk through the procedure to the end." Even a frontier model can fail a simple constraint like "don't skip any endpoint." The 100% in the title means the contract can force the model to walk the procedure . CoT cannot be inspected if you leave it as free prose. The real question isn't "how long does the model think" — it's "can we turn that thinking into a submittable audit artifact?" The focus shifts from co