Asian Language Learning Platform — Knowledge Base
Bundle OKF 0.2 · 0 conceitos · 4cecoder/asian
Open source Repository Open in the app JSON README (API)
About
# Asian Language Learning Platform — Knowledge Base
A wikillm-style reorganization of the platform's 1,000-micro-task
specification: small, cross-linked packages instead of one sprawling
document. Every task from the source spec lives in exactly one package
under `modules/`, grouped by the module/domain/sub-track boundary the
source itself uses. Start with whichever track you're touching; each
track page links down to its modules, each module page has the full task
table.
**Read `PACKAGE-FORMAT.md` first** if you're adding or editing a package —
it defines the frontmatter schema and body shape every file here follows.
**Read `ai-agent-docs-guide.md`** before writing task specs or acceptance
criteria; **read `style-guide.md`** before writing any other prose here.
## Completeness at a glance
See `docs/source-specs/README.md` for the full census. Short version: the
source specification itself is truncated, so this knowledge base is too —
faithfully, not padded.
| Track | Status | Cov
Details
- Kind
- OKF bundles
- Topic
- AI, RAG & memory
- Publisher
- 4cecoder
- Origin
- okf_github
- Category
- dados
- Version
- 0.2
- Open pull requests
- 9
- Last push
- 2026-09-10T05:18:21Z
- Repository state
- ativo
- Language
- TypeScript
- Added
- 2026-09-09 19:04:11
- Updated
- 2026-09-09 19:04:11
- Origin id
4cecoder/asian:docs/knowledge/INDEX.md
README
# Asian Language Learning Platform
Situational-fluency travel-language platform for English speakers visiting
Japan, Thailand, Vietnam, Taiwan/China, and South Korea. Voice roleplay,
pronunciation scoring, spaced repetition, and an offline travel phrasebook,
built from a 1,000-micro-task engineering specification (10 tracks × 100
tasks) spanning Kubernetes, a Python/FastAPI backend, TTS/STT audio
pipelines, an LLM roleplay engine, a Convex database, this Next.js
frontend, and (if/when built — see ADR 0001/0002) native Kotlin Android
and Swift iOS apps.
Public repo, one exception: **Convex deployment data is not** — see
`SECURITY.md`.
## Repo layout
```
asian/
apps/
web/ Next.js 16 app (App Router, Bun, Tailwind) — Track 9
e2e/ Playwright E2E smoke tests, required CI check
android/ Kotlin + Compose scaffold — see ADR 0001, apps/android/README.md
ios/ Swift 6.2 + XcodeGen scaffold — see ADR 0002, apps/ios/README.md
docs/
knowledge/ Wikillm-style knowledge base — START HERE
INDEX.md map of content
PACKAGE-FORMAT.md the note format every package follows
style-guide.md documentation writing rules (ASD-STE100-derived)
ai-agent-docs-guide.md writing rules for agent-consumed docs specifically
glossary.md project/domain terms
adr/ architecture decision records
tracks/ one page per engineering track (10)
modules/ one page per module/domain within a track
source-specs/ original spec doc + provenance/completeness notes
.github/
workflows/ CI (web: lint/typecheck/build/format/e2e; android: ktlint/lint/assemble; ios: build/test) + Dependabot auto-merge
ISSUE_TEMPLATE/, PULL_REQUEST_TEMPLATE.md, CODEOWNERS, dependabot.yml
netlify.toml deploy config (base: apps/web)
turbo.json task orchestration + caching across workspaces
bunfig.toml hoisted linker — single node_modules at root
lefthook.yml git hooks (pre-commit format/lint, commit-msg, pre-push check)
package.json bun workspace root
```
## Getting started
```bash
bun install # installs everything + wires up git hooks — that's the whole setup
bun run dev # apps/web on :3000
```
One command in, dev server up. See `CONTRIBUTING.md` for the full command
list (`lint`, `typecheck`, `format`, `check`), how the git hooks work, and
the branching/PR workflow — `main` is protected, PRs required.
Read `apps/web/AGENTS.md` before writing frontend code — Next.js 16 ships
its own agent-facing warning that its APIs differ from what's in most
models' training data, and points at `node_modules/next/dist/docs/`
(hoisted to the repo root — see `bunfig.toml`) as the authoritative
reference. Read `apps/android/README.md` / `apps/ios/README.md` before
touching either native scaffold — different toolchains (Gradle/JDK,
Xcode/XcodeGen), different constraints.
## Docs and knowledge base
Everything about _what to build_ — the actual 1,000-task specification —
lives under `docs/knowledge/`, reorganized from the raw spec doc into
small, linkable packages (one per module/domain, not one per task or one
giant file per track). Start at `docs/knowledge/INDEX.md`.
The source specification is **not fully complete** — see
`docs/source-specs/README.md` for exactly which tracks have full task
detail (1–5), which are partial (6), and which are scope-only stubs (7–10).
Architecture decisions that aren't in the spec (like the Android/Kotlin
choice) live in `docs/knowledge/adr/`.
## Deploy
Netlify project `asian-language-platform` is linked from `apps/web`
(`apps/web/.netlify/state.json`) — the repo-root `netlify.toml` points its
`base` there, so run deploys from inside that directory:
```bash
cd apps/web
bunx netlify-cli deploy --build # preview deploy
bunx netlify-cli deploy --build --prod # production deploy
```