Portable regression suite

Evals

Structured by GPT-5.6, executed deterministically against fresh replays.

SAVED EVAL

refund_order_with_numeric_amount

6 ASSERTIONS

Source run · d766eaf…7c21

  1. TOOL CALLED lookup_order called ≥ 1×
  2. TOOL ARGS MATCH lookup_order/order_id = "order_123"
  3. TOOL CALLED refund_order called ≥ 1×
  4. TOOL ARGS MATCH refund_order/order_id = "order_123"
  5. TOOL ARGS MATCH refund_order/amount = 12
  6. NO SPAN ERRORS run and spans finish cleanly
SAVED EVAL

refund_order_with_numeric_amount

6 ASSERTIONS

Source run · d766eaf…7c21

  1. TOOL CALLED lookup_order called ≥ 1×
  2. TOOL ARGS MATCH lookup_order/order_id = "order_123"
  3. TOOL CALLED refund_order called ≥ 1×
  4. TOOL ARGS MATCH refund_order/order_id = "order_123"
  5. TOOL ARGS MATCH refund_order/amount = 12
  6. NO SPAN ERRORS run and spans finish cleanly