17 releases 20 tags

RSS feed

• v9.3.0 c69897c55b

  Compare

  v9.3.0 Stable

  claude-bot released this 2026-07-18 17:11:27 +00:00 | 40 commits to main since this release

  Three self-contained v10-backlog research-workflow utilities.

  OCR — image text into searchable content (#118)

  On image upload the configured vision model also transcribes visible text (lyrics, captions, on-screen text). It's offered on the media card — "Show extracted text (OCR)" → "Add transcribed text to entry" — and only appended to the entry's content on explicit accept (sanitized on write), at which point it's full-text searchable. Nothing is applied automatically; no-ops when AI isn't configured. Migration 055 adds media.ocr_text.

  Mastodon + Instagram import adapters (#121)

  Two new archive parsers behind the existing adapter interface: Mastodon outbox.json (ActivityPub — original posts only, boosts skipped, HTML stripped) and Instagram posts JSON (caption + creation_timestamp), each mapped to entries with the right source/source_id dedup shape. New /api/import/{mastodon,instagram} endpoints + Import-page cards.

  Offline read cache (#122)

  Recently-viewed entries and motifs are now readable offline via an app-managed, user-scoped IndexedDB cache (separate from the capture queue). When a fetch fails offline the detail page renders the cached copy with a banner; the cache clears on logout. This does not relax the invariant that the service worker never caches /api/*.

  ---

  API suite 51 files / 368 tests · app suite 14 files / 76 tests · builds green.

  Downloads

  • Source code (ZIP)
    0 downloads
  • Source code (TAR.GZ)
    0 downloads