VidSavant - local-first video knowledge workspace

VidSavant is the successor to the earlier VideoLearningTool prototype: a local-first desktop app that turns a video URL into a searchable, transcript-grounded knowledge item with template summaries, timestamp-cited chat, optional downloads, and durable background jobs. By 2026-07-21, the implementation had completed Phase 32 of the upgrade roadmap: scoped chat now supports immutable video/tag/selection/library snapshots, bounded cross-video FTS evidence, validated [V MM:SS] citations with workspace deep links, degraded-member handling, explicit preflight/consent, restart-safe sessions, and snapshot-aware Markdown export. Phase 32 verification passed 805 tests (53 deselected), 51 Chromium browser tests, and ruff. The project remains personal and local-first; no live provider run was made for this phase.