A cone that looks wrong took a screenshot and a paragraph to describe.
"Copy sightline report" on any view cone's context menu now puts the
cone's own inputs and results on the clipboard: the pose, the standing
level it chose, every support under the observer, and 25 rays across the
aperture with the wall each one stopped on and that wall's bands. The
same JSON and a 240-unit crop of the model around the origin land in
<app support>/sightline-reports, written after the toast so nothing
waits on the disk. Maps still on the legacy geometry report their pose
and say so instead of going quiet.

The builder takes the exact inputs the cone was drawn from and recasts
them, so a report always describes the cone on screen.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>