IT Passport Learning
A trilingual past-exam practice app for the IT Passport (ITパスポート) certification exam
29 IPA past exams (FY2009–FY2026), a syllabus-aligned textbook and a glossary — in JP / ZH / EN.
:gb: English · :cn: 中文版 README
What is this?
itlearn.bojiangz.com — ITパスポート 過去問ノート, a study app for the Japanese IT Passport exam built around the real past questions.
Non-native learners of Japanese technical exams get blocked by kana/kanji recognition, not by concepts. アクセシビリティ → accessibility → 可访问性, seen once, sticks. So every question keeps its original Japanese text with a Chinese and English translation you can show or hide, and every explanation is available in all three languages.
What's in it
| Page | What it does |
|---|---|
| Today | Daily plan: mistakes to redo, reviews that are due and new questions (picked by the exam's field ratio). New users start with a 20-question baseline. Also shows progress over the 2,900 past questions, an exam-date countdown, weak topics (low accuracy × often tested), accuracy by field against the pass rule (600/1000 overall, ≥300/1000 in each field) and 3 glossary terms for today. |
| Practice | One question per screen: browse by exam or by topic, practise with your own settings, or redo your mistakes. Japanese question text with a ZH/EN translation toggle, enlargeable figures, and a trilingual explanation after you answer. |
| Mock exam | 100 questions in 120 minutes, as in the real exam: a random draw in the official field ratio (strategy 35 / management 20 / technology 45) or one whole past exam. Scored 0–1000; results only at the end. |
| Textbook | Syllabus-aligned trilingual study units — 63 topics, 244 units — with key points, figures and check questions taken from the past exams. |
| Glossary | 1,395 textbook terms, searchable in Japanese, Chinese and English, each linked to its textbook unit and related past questions. |
| Record | Totals, reviews due over the next 7 days, questions per day (5 weeks) and accuracy by topic. |
| Settings | Exam date, daily review limit, daily new questions, default translation visibility, interface language. |
The interface language switches between 日 / 中 / EN. Your learning record is kept in the browser on your device; no account is needed.
AI tutor (bring-your-own-key preview)
An AI tutor (AI 先生) lets you ask about a question, a term or a textbook unit, and get a review of your weak topics. It is a preview that needs your own Anthropic API key: the key stays in your browser, which calls Anthropic directly. Open it from Ask AI Tutor after answering a question or on a glossary term, or from Settings; it is not in the main navigation yet. A full release is planned.
Quick start
Visit itlearn.bojiangz.com — no install needed.
Architecture
itlearn.bojiangz.com
│
▼
┌──────────────────────────────────────────┐
│ Next.js 15 (App Router, Turbopack) │
│ React 19 · Tailwind CSS v4 · next-intl │
├──────────────────────────────────────────┤
│ Pages /[locale]/… │
│ today (home) · practice · mock │
│ textbook · glossary · record │
│ settings · quiz (old-link redirect) │
├──────────────────────────────────────────┤
│ API routes │
│ /api/quiz/questions │
│ /api/quiz/feedback (owner only) │
│ /api/learning (owner sync) │
│ /api/sensei/terms (tutor lookup) │
├──────────────────────────────────────────┤
│ Data: committed JSON corpus │
│ data/ip/quiz 2,900 questions │
│ data/ip/textbook 244 units │
│ Learning record: browser storage │
│ (+ Upstash Redis sync for the owner) │
├──────────────────────────────────────────┤
│ Vercel · Analytics · Speed Insights │
└──────────────────────────────────────────┘
▲
│ built offline by
scripts/ (Node / TypeScript pipeline)
IPA past exams + syllabus → Claude extraction,
checks and translation → trilingual corpus
Content
| Item | Source | What the app has |
|---|---|---|
| Past questions | IPA 過去問題, FY2009–FY2026 (29 exams) | 2,900 questions, 511 with figures; Japanese original + ZH/EN translation; trilingual explanations; each question mapped to one of 63 syllabus topics |
| Textbook | IPA シラバス Ver.6.5 as the skeleton, AI-generated | 244 units in 63 topics (strategy 98 / management 23 / technology 123), 1,417 term entries |
| Glossary | Built from the textbook terms | 1,395 entries |
The past-exam text was extracted from the IPA PDFs with Claude vision, then repaired and checked against the source in several full passes, with independent sample audits recorded under evidence/.
Status
| Stage | Status | What it covers |
|---|---|---|
| Phase 1–4 (first app) | :white_check_mark: Retired | First version built on a purchased textbook (Book / Quiz / Glossary / Chat / Tutor, three theme shells). Replaced by the IPA-based rebuild below. |
| Stages 1–3.5 — IPA source data | :white_check_mark: Done | Syllabus structure; all 2,900 past questions extracted, repaired and mapped to syllabus topics. |
| Stage 4 — Textbook v1 | :white_check_mark: Done | 244 trilingual units generated. |
| Stage 6 — App data integration | :white_check_mark: Shipped | Textbook reader, past-question practice with translations and explanations, glossary. |
| Stage 7 — UI redesign | :white_check_mark: Shipped | One design for every page (Today / Practice / Textbook / Glossary / Record / Settings), mock exam, learning record, new textbook reader. |
| Stage 7 — Textbook v2 batch rewrite | :hourglass: In progress | Units are being rewritten to the v2 format in batches: 13 of 244 so far. |
| AI tutor | :test_tube: Preview | Bring your own Anthropic key; reachable from Ask AI Tutor and Settings. Full release planned. |
The live state is in docs/STATE.md.
Tech stack
| Layer | Technology |
|---|---|
| Framework | Next.js 15.5 (App Router, Turbopack) |
| UI | React 19, Tailwind CSS v4, Radix UI, Lucide icons |
| i18n | next-intl (Japanese / Chinese / English) |
| Storage | Browser storage for the learning record; Upstash Redis for owner sync and question feedback |
| Testing | Vitest (1,264 tests), Playwright (e2e) |
| Deploy | Vercel, Vercel Analytics, Speed Insights |
| Content pipeline | Node / TypeScript scripts; Claude (Anthropic) for extraction, review and translation |
Repository layout
.
├── apps/web/ Next.js 15 web application (the only pnpm workspace member)
│ ├── src/app/[locale]/ Page routes (practice, mock, textbook, glossary, record, settings, …)
│ ├── src/app/api/ API routes (quiz, learning, sensei)
│ ├── src/components/ UI components
│ ├── src/lib/ Data loaders, learning record, review scheduling, mock exam
│ ├── public/quiz-figures/ Past-exam figures (WebP)
│ ├── e2e/ Playwright tests
│ └── messages/ i18n strings (ja/zh/en)
├── data/ip/quiz/ Committed question corpus (questions, translations, explanations)
├── data/ip/textbook/ Committed textbook units, figures, unit index
├── scripts/ Extraction / repair / generation pipeline scripts
├── docs/
│ ├── STATE.md Live project state (truth source)
│ ├── decisions/ ADRs for major decisions
│ └── discussion/ Session logs
├── evidence/ Audit evidence (Rule A)
└── failures/ Failed attempt archive (Rule B)
The raw IPA PDFs and intermediate data (data/ip/exams, data/ip/sources, data/ip/syllabus, .source/) are gitignored.
Development
Requires Node ≥ 20.9 and pnpm 11.
# Clone
git clone https://github.com/hakupao/it-passport-learning
cd it-passport-learning
# Install (from the workspace root)
pnpm install
# Web app
cd apps/web
pnpm dev # http://localhost:3000
pnpm build
# Checks
pnpm test # Vitest
pnpm lint
pnpm test:e2e # Playwright e2e
# Quiz answer-key cross-check (repo root)
cd ../..
pnpm check:quiz:ci # committed data only
The app reads the committed corpus under data/ip/, so it runs locally without any environment variables. Without Redis and an owner key, the learning record simply stays in the browser.
Environment variables
All optional.
| Variable | Description |
|---|---|
OWNER_KEY |
Site-owner secret. Opening any page once with ?owner_key=<OWNER_KEY> sets the owner cookie, which enables record sync and question feedback. Unset = nobody is owner. |
UPSTASH_REDIS_REST_URL / UPSTASH_REDIS_REST_TOKEN |
Upstash Redis for the owner's record sync and question feedback (KV_REST_API_URL / KV_REST_API_TOKEN are also accepted). |
QUIZ_DATA_ROOT / TEXTBOOK_DATA_ROOT |
Override the data directory (default ../../data/ip, relative to apps/web). |
Project statistics
| Metric | Value |
|---|---|
| Past questions | 2,900 (29 exams, FY2009–FY2026) |
| Textbook | 63 topics, 244 units (13 on v2) |
| Glossary | 1,395 entries |
| Locked decisions | 203 (D-001 … D-203) |
| Session logs | 73 in docs/discussion/ |
| Unit tests | 1,264 (Vitest, 110 files) |
License
- Code, pipeline, ADRs — License pending (consult the repo owner before redistribution).
- Past-exam questions — Based on the past exam questions published by IPA; copyright belongs to IPA. The text has been OCR-extracted, cleaned, classified by syllabus topic and translated, so it differs from the original.
- Textbook and explanations — AI-generated for this project.
Contributing
This is a personal R&D project. Issues and PRs welcome. Read docs/STATE.md and the most recent session log before opening anything substantive.
