Good news on the index: no duplicate pages — only lowercase canonical slugs are stored, mixed-case variants don't resolve. So the index is clean. The remaining cleanup is purely in git: 20+ tracked meeting paths still carry uppercase names (the old files), which resolve to the same inodes as lowercase but keep the case-collision landmine in the index. Since

core.ignorecase=true

, a plain

git mv

won't work; I'll force-rename them in the index to canonical lowercase. Let me do it safely.