fix(email): eval-pass follow-ups — ctx-stamped baseline, refresh limit, usage aggregate#2089

Merged

itomek merged 3 commits into

mainfrom

tmi/eval-pass-followups

Jul 15, 2026

Commits on Jul 15, 2026

test(email): re-record baseline_accuracy.json at the pinned 16K window (#1892)
Tomasz Iniewicz
committed
ci(email): cut scorecard-refresh limit 60->25 to fit the pinned 16K ctx (#2087)
Tomasz Iniewicz
committed
fix(eval): surface triage token usage in the aggregate scorecard (#1891)
Tomasz Iniewicz
committed