Synapse is a study workspace. Organize courses read and annotate the material, train with generated exercises and watch your progress forked from denishovart.com/synapse
6 folders · 12 files

e2e
messages
project.inlang
src
static
web-client
.dependency-cruiser.cjs
.env.web-client
index.html
package.json
playwright.config.ts
README.md
stryker.conf.json
svelte.config.js
tsconfig.json
vite.config.js
vitest.config.ts
vitest.setup.ts
README.md

@synapse/web #

The Synapse UI, a single-page app in Svelte 5 (runes) built with Vite. index.html mounts src/app/main.ts, which mounts App.svelte, which renders the active page from a client router over the History API. One build runs everywhere. The NestJS API serves it, the Tauri shells embed it, and it also deploys on its own as a static site with nothing behind it.

src has three layers and imports flow one way, from lib to pages to app; dependency-cruiser fails pnpm lint if that is broken. app/ is the frame: boot, router, top bar, auth gate. pages/ is one folder per screen, nested the way the navigation is, so the viewer, the split-PDF wizard, the exercises and the per-course settings sit under pages/course/. lib/ is the shared infrastructure with no screen of its own. The Loro documents and their IndexedDB persistence sit in data/, the HTTP, LAN and file-pack sync engines in sync/, the upload and export pipelines in documents/, then the Tiptap editor, the AI client and the Paraglide message catalogs. pdf.js is a bundled dependency loaded on demand, so the viewer behaves the same in dev, in the built app and in the shells.

The workspace on the device is the source of truth, which is why pages call the repositories in lib/data/repos/ and never fetch study data over HTTP. Adding a server endpoint for something a document field could carry is almost always the wrong move. Read Architecture before changing the data layer.

The scripts, the tests and the i18n rules are in Development.