The July 28 sync is now fully in the brain and readable — content, summary, all of it. (The

grep -c

returned 0 only because

list

truncates its output, not because the pages are missing; the direct

get

proves they're indexed.) Posting the summary to Royce: Good catch — you were right, I was stuck at July 21. Root cause + fix: What was broken: the Granola ingest job was pulling the notes fine and writing the

.md

files to

~/company-brain/meetings/

(July 24, 27, 28 all on disk + committed to git), but the

gbrain sync

step was silently failing. It bookmarked those files as "already imported" without ever putting them in the index — so they existed as files but were invisible to `query`/`get`. Underlying trigger: case-collision slugs on this case-insensitive Mac filesystem (e.g.

...eVr6op

vs

...evr6op

) were wedging the incremental sync, and some earlier git-commit crashes advanced the bookmark past unimported files. Fix: forced a hard reimport of the meetings dir → 157 pages imported, 235 chunks embedded. July 24/27/28 syncs (plus the 07-28 db-rollback discussion) are now queryable. Confirmed I can pull the July 28 sync content directly. Two follow-ups I'd flag: (1) the ingest script's commit/sync step should fail loudly instead of silently advancing its bookmark, and (2) the duplicate case-collision pages should get cleaned up. Both are in the ingest script — want me to draft fixes for you to review, or leave those jobs alone per the "don't touch ingest unless Royce asks" rule? And yes — now that I actually have the July 28 sync, want me to answer your earlier question about this week's priorities based on it?