The first replacement failed its comparison.
Eli was carrying a box of unpowered reference pieces into the workroom when the independent operator pushed back from the table. The new device record did not match the retained physical mark at her site. A technician asked her to repeat the value through the old receiver so he could see where the difference began.
“No,” she said. “That puts the new credential into the channel we’re replacing.”
The technician opened his mouth, stopped, and turned to Juno.
“She is right,” Juno said.
The operator’s suspicion had not identified sabotage. It had identified a difference. They checked the local component description through the trusted contact, compared the dated installation record, and found a swapped unit label from an earlier maintenance visit. The physical piece had not been secretly altered. Its record had been copied into the wrong row. Juno made the correction visible to the site operator and kept the failed comparison with the replacement history.
The test window continued to shrink while they did it.
A volunteer who had traveled in for the afternoon comparison sat on the floor beside his coat. He had a return escort available for only another hour. He asked whether he should wait or go.
