← Roadmap

#10 · Corpus · 2026-05-10

Local corpus materialization and import UX

Status: Backlog

Corpus

Implementation

  • Persist the heavy corpus locally per device (IndexedDB or equivalent) after user-confirmed import.
  • Show progress and cancellable/restartable behavior; first-time and “new device” paths share core pipeline where possible.
  • Align with spec—bulk text does not sync device-to-device through the cloud.

Questions

  • Storage quota handling and cleanup when switching upstream repo?
  • Whether partial import is allowed or all-or-nothing for consistency?