permission_history writes two 'decided' rows for one cross-box notify decision (200 ms apart)

by adom

Adom Bridge unleashes your AI onto your full PC: full power, full safety. The AI breaks out of the container it runs in and onto your real machine, managing and running your entire laptop as you: launch apps, move files, drive any window, control KiCad, Fusion 360 or your real signed-in browser. Works with any AI, cloud or local (Claude Desktop, Claude Code, Codex), no Hydrogen required. Bridge and its bridges are an operating system for AI, with managed Node and Python runtimes, and a human-onl

Overview Files Skills Discover Packages Releases Dependencies Issues19 Breadcrumbs Pull Requests1 Activity

Closed bug report

John Lauer · 1mo ago ·closed by John Lauer

Seen on ConfRoomROG 2.1.141 (2026-09-21 00:43Z): a buttoned notify_user prompt decided from AdomLapper produced two 'decided' rows for the same id, decision Yes, decidedOn AdomLapper, decidedAt 00:43:44.074Z and 00:43:44.265Z. One decision, one row: the resolve path (perm_mirror_resolve -> resolve_permission_decision_from -> dismiss_from) and the notify teardown (forget_notify_decision + permission_mirror::dismiss) both record. Dedupe at record_decision: skip when the last row for that id is a decision within a few seconds with the same decision, or have the second caller use refan_dismiss. Cosmetic for readers that dedupe by id; wrong for a count.

2 Replies

John Lauer · 1mo ago

Fixed in 2.1.143 (building): record_decision dedupes by id over a 30 s window, so the resolve path and the notify teardown recording the same decision 200 ms apart write one row. Proof (a cross-box decision, then permission_history showing exactly one decided row for that id) when it is on the fleet.

John Lauer · 1mo ago

Fixed in 2.1.143: record_decision remembers the ids decided in the last 30 s and skips a repeat, so the resolve path and the notify teardown recording the same decision ~200 ms apart write one row.

Proof on arav-rog (2.1.143), the same shape as the report: a buttoned notify_user raised on arav-rog, mirrored to winvm, decided there through the card path (ui_decide on winvm's loopback, the click path), applied on arav-rog:

origin notify_user -> action="Yes" timedOut=false id=09cc184934414fd7
arav-rog permission_history rows for 09cc184934414fd7: 1
  -> ('decided', 'Yes', decidedOn 'winvm', '2026-09-21T13:37:48.002Z')

(The timeout path also leaves exactly one row: five earlier attempts that lapsed each show a single ('decided','timeout','arav-rog',...).) Rows written before 2.1.143 keep their duplicates; a reader that counts should dedupe by id for history older than today.

John Lauer closed this · 1mo ago