- The study feed's progress header and bottom pager no longer use the old
navy glass: new .study-chrome class uses the neutral surface token
(same gray as the cards, #2f2f2f in dark), the neutral border
(8% white in dark, #e5e5e5 in light) and a neutral shadow.
- Close tile, streak pill and XP pill now use --surface-sunken /
--text-muted, matching the app's other buttons and chips; the close
glyph is the shared XIcon.
- Bottom pager: counter in --text, arrows in --text-muted.
- .glass (mobile header) no longer has the hardcoded #101720 navy; it is
now color-mix of the surface token. Removed the unused --color-night-*
tokens and the last night- usage (create screen sticky bar).
- Hidden the 'Swipe up' hint on desktop, where it collided with the
bottom pager.
New deck-chrome suite: 15/15 (computed surfaces/borders/text in both
themes, no navy); pwa 16/16, final-ux 12/12, security 8/8, contrast 20/20.
Next.js 15 app that turns photos, PDFs, DOCX/PPTX and text into a swipeable
feed of AI-generated activities (flashcards, multiple choice, true/false,
fill-in-the-blank, matching, ordering, short answer) using the user's own
Ollama models.
Highlights:
- All AI calls run in server route handlers; the browser never talks to Ollama.
- Bring-your-own-key mode (default): the key lives only in the browser and is
sent per-request; server, local-daemon and shared-key modes also supported.
- Client-side document extraction (pdf.js, mammoth, jszip) with scanned-PDF
page rendering for the vision model; notes stay in their original language.
- ChatGPT-style shell: collapsible sidebar / mobile drawer, neutral token-based
light + dark themes, folder organisation, PWA (manifest, SW, offline page).
- Strict JSON validation with zod + auto-repair, rate limiting, and request
guards (cross-origin, body-size, API-key sanitising).
- setup.sh bootstraps deps, .env.local and a production build.
Gitignore excludes .env*, the generated pdf.js worker and local scratch files.
