LeetCode Tracker

A personal LeetCode practice app with bilingual problem content, curated study lists, company tags, progress tracking, AI-generated explanations, and a local online judge.

The project is a TypeScript monorepo:

• client/ - React 19 SPA, Vite, Tailwind CSS v4, shadcn/ui-style components, wouter routing.
• server/ - Express, tRPC, scheduled sync endpoints, OAuth/session plumbing, judge runtime.
• shared/ - Zod schemas and shared types used by both client and server.
• drizzle/ - MySQL schema and migrations.

Explore the engineering

Start here if you are reviewing the project rather than setting it up:

Area	Implementation	Checks
Data model	Drizzle schema	Schema tests
Sync orchestration	Task registration and run tracking	Orchestrator tests
Bilingual ingestion	Problem ingestion · Translation	Detail sync tests
Local judge	Runner · SQL judge	Judge tests
Delivery	GitHub Actions workflow	Latest CI runs

Deployment boundary: the judge runs submitted code in local subprocesses. Time and memory limits do not make it a hardened sandbox for untrusted public users. See Architecture Notes before deploying.

Current Status

• GitHub repo: teddy8023mars/leetcode-tracker
• Default branch: master
• CI: GitHub Actions runs pnpm check and pnpm test
• Current check results: GitHub Actions
• Main local workflow: use git worktrees for new feature work

Features

• Problem list with search, difficulty/status/company/topic filtering, and pagination.
• Problem detail page with bilingual content fallback, related problems, tags, company labels, official/AI solutions, and a side-by-side solve panel.
• Local code editor powered by Monaco with Python/Java/C++ templates.
• Online judge pipeline with generated test suites, cached submissions, verdicts, and first-failing-case details.
• User progress tracking with todo/reviewing/done states and SM-2 spaced repetition fields.
• Sync pipeline for LeetCode study plans, LeetCode detail data, Chinese content fallback, and company tags.
• Owner-only manual sync controls and heartbeat-protected scheduled endpoints.

Quick Start

Requirements:

• Node.js 24
• pnpm 10.4.1
• MySQL if you want real persisted data

pnpm install
cp .env.example .env
pnpm dev

The dev server starts Express and Vite together. By default it tries port 3000 and will use the next available port if needed.

Without DATABASE_URL, the server still boots, but most data-backed screens return empty data. That is useful for frontend-only work, but real sync/progress/judge history requires a database.

Environment

Do not commit .env. Start from .env.example.

Important variables:

• DATABASE_URL - MySQL connection string used by Drizzle.
• JWT_SECRET - session cookie signing secret.
• OWNER_OPEN_ID - enables owner-only sync actions for that OAuth user.
• OAUTH_SERVER_URL, VITE_OAUTH_PORTAL_URL, VITE_APP_ID - Manus OAuth/app wiring.
• HEARTBEAT_SECRET - protects /api/scheduled/* endpoints.
• BUILT_IN_FORGE_API_URL, BUILT_IN_FORGE_API_KEY - OpenAI-compatible Forge endpoint used for translation and test generation.
• BOOTSTRAP_SKIP_CN, BOOTSTRAP_SKIP_LLM - optional bootstrap switches for long-running sync jobs.

Common Commands

pnpm dev       # Run Express + Vite dev server
pnpm build     # Build client and bundle server into dist/
pnpm start     # Run production build
pnpm check     # TypeScript type check
pnpm test      # Vitest test suite
pnpm format    # Prettier write
pnpm db:push   # Generate and run Drizzle migrations

Single test example:

pnpm vitest run server/__tests__/sync.leetcode.detail.test.ts

Development Workflow

Use a new worktree for every feature or cleanup:

git fetch origin
git worktree add /private/tmp/leetcode-tracker-my-feature -b codex/my-feature origin/master
cd /private/tmp/leetcode-tracker-my-feature
pnpm install --frozen-lockfile

Before pushing:

pnpm check
pnpm test
git diff

Keep commits small and behavior-focused. Good commit examples:

• fix: stabilize scheduled sync tests
• feat: add review dashboard
• docs: add project setup guide

Architecture Notes

API

The app exposes a typed tRPC API at /api/trpc. The root router is assembled in server/routers.ts and currently includes:

• problems - list/detail/solutions/company tags
• lists and companies - metadata for curated lists and company tags
• progress - local progress and spaced repetition
• judge - authenticated code submission and submission history
• aiSolutions - cached AI explanations and admin generation
• sync - sync status and owner-only manual triggers
• auth and system - session and system plumbing

Data

The primary schema lives in drizzle/schema.ts. Tests use an in-memory SQLite helper in server/testHelpers/inMemoryDb.ts, so schema changes usually need a matching test DDL update.

Sync

Sync jobs are registered in server/sync/index.ts and executed through the orchestrator in server/sync/orchestrator.ts. Scheduled HTTP endpoints live in server/scheduled.ts and are protected by HEARTBEAT_SECRET.

Judge

The judge writes temporary files, runs user code in subprocesses with time and memory limits, and stores submissions. It is suitable for local/dev use; a real production deployment should replace it with a stronger sandbox.

Documentation

• Developer log
• M1 verification report
• Planning/spec history under docs/superpowers/

Notes for Future Work

• Prefer GitHub issues for new features before implementation.
• Keep .env.example updated when adding environment variables.
• Run pnpm check and pnpm test before every push.
• For database schema changes, update Drizzle schema, migrations, and in-memory test DDL together.