minormonitoring-gapverifiedfirsthand
Phone captures kept landing in a second, unmanaged Obsidian vault next to the real one
Posts and articles saved from a phone's share sheet were going into a second, unmanaged Obsidian vault sitting next to the managed one. It was discovered on September 15, 2026, and 64 items were recovered.
Cause
The same iCloud Obsidian folder held a second vault alongside the managed one. The phone's share sheet always saved there, the owner just thought of it as the "Inbox", and the Mac's Obsidian app had never opened either of the two vaults.
Consequence
Everything captured on the phone sat somewhere it was never organized or searched. Nothing was lost, but none of it was visible in the knowledge base.
Fix
The 64 items were archived immutably and turned into 61 pages (two skipped as a body-less bookmark and an empty file, one merged into an existing page). Along the way, 31 Unicode normalization violations and one filename containing a newline were fixed.
What happened
When something interesting turned up on the phone — a post, an article — it went into Obsidian through the share sheet. Later it would be organized into the knowledge base on the Mac. That was the plan.
On September 15, 2026, it turned out the share sheet had never been sending anything to the managed vault. Inside the same iCloud Obsidian folder, right next to the managed vault, was a second vault. That’s where the phone had been saving everything.
The chaos on the ground
The reason nobody noticed is that nothing could see it. The owner called that destination the “Inbox” and assumed captures were landing in a proper inbox. Meanwhile, the Obsidian app on the Mac had only ever opened a backup copy on an external drive; neither of the two iCloud vaults was even registered in the app. The tool that managed the knowledge base only looked at one vault, and nothing reported what was piling up next door.
Every save “succeeded.” Not a single error. The captures were simply going to the wrong place.
It wasn’t the first time, either. When the knowledge base was first built, it was set up on the backup copy instead of the live vault and had to be moved afterward. With several vaults of similar names in similar places, “which one is the real one” gets fuzzy fast.
Root cause
The capture side and the organizing side were looking at different vaults. The phone’s share sheet saved to the second vault, while the knowledge-base tooling only handled the first. Nothing connected the two, and nothing flagged that notes were accumulating somewhere unmanaged.
The fix
The 64 items in the second vault were first captured as an immutable archive, then 61 of them were turned into pages. Two were skipped — a bookmark with no body and an empty file — and one was merged into an existing page. Four new topic index pages were added and nine existing ones updated, taking the organized sources from 277 to 338.
Two more problems surfaced during the import. Thirty-one filenames from the phone were in Unicode normalization form NFD, and one filename contained a newline. Because APFS treats names that differ only in normalization as the same name, they couldn’t be renamed in place; each had to go through a temporary name first.
The lesson is simple: check that the place you capture into is the same place you organize from. “Saved successfully” doesn’t mean “saved somewhere that’s managed.”
Sources
- Firsthand account from the site operator. There is no public write-up to link to.