🧩 OpenSolve — The Open-Source LeetCode Alternative

OpenSolve is a next-generation, open-source platform for algorithmic practice, coding challenges, and technical interviews — built for the community, by the community. It’s a self-hostable, extensible, and beautifully designed clone of LeetCode with full-stack flexibility and zero paywalls.

---

🚀 Why OpenSolve?

LeetCode is great — but closed. OpenSolve brings the same experience to the open world:

• 💻 Open Source — 100% free to use, self-host, and modify.
• ⚙️ Modern Stack — Built with Next.js, PostgreSQL, and TypeScript.
• 🔐 Self-Hostable — Deploy locally or on your own cloud (Docker support included).
• 🧠 Smart Judge — Supports multi-language execution (Python, C++, Java, JS) with sandboxed Docker containers.
• 🏗️ Problem Builder — Create, import, and manage your own problems with test cases and constraints.
• 🧑‍💻 User Profiles & Leaderboards — Compete, compare, and climb the ranks.
• 🎨 Sleek UI — Clean, minimal, and dark-mode ready.

---

🧱 Tech Stack

Layer	Technology
Frontend	Next.js, TailwindCSS, React Query
Backend	Next.js, PostgreSQL, Prisma ORM
Judging System	Docker-isolated code runner (supports multiple languages)
Auth	JWT + OAuth2 (GitHub, Google)
Deployment	Docker Compose / Fly.io / Railway / Supabase backend
AI Features (optional)	OpenAI / Ollama local inference for hints, explanations, and code optimization

---

🧠 Roadmap

• Problem CRUD & submissions
• Code execution engine
• Contests & rating system
• Teams & collaborative problem solving
• Offline mode for universities & bootcamps

---

🤝 Contributing

We love PRs! Clone, fork, or self-host — and help make OpenSolve the ultimate open coding playground.

git clone https://github.com/opensolve/opensolve.git
cd opensolve
docker compose up --build

Prefer shortcuts? Use make up or npm run docker:up. The default stack starts the web app, Postgres, RabbitMQ, MinIO, and the judge worker.

Useful local URLs:

• App: http://localhost:3000
• RabbitMQ: http://localhost:15672 (opensolve / opensolve)
• MinIO console: http://localhost:9001 (opensolve / opensolve-secret)

To use a private local env file, copy ops/env/dev.env and point Compose at it:

cp ops/env/dev.env ops/env/dev.local.env
OPENSOLVE_ENV_FILE=./ops/env/dev.local.env docker compose up --build

---

⚡ License

MIT License — Free for personal, educational, and commercial use. No restrictions. Just code.

---

🌍 Vision

“Code is meant to be shared, not locked behind paywalls.” OpenSolve aims to democratize algorithmic practice and make technical learning accessible to everyone — from students in Kabul to developers in Silicon Valley.

---

🪣 Object Storage (MinIO / S3)

Uploads (avatars, attachments, future problem assets) use the S3 API via lib/storage/minio.ts. For local/self-hosted deployments we bundle a MinIO stack that behaves exactly like AWS S3.

1. Set the environment variables (see .env.example for defaults):

   MINIO_ENDPOINT=localhost:9000
   MINIO_BUCKET=opensolve-assets
   MINIO_ACCESS_KEY=opensolve
   MINIO_SECRET_KEY=opensolve-secret
   MINIO_REGION=us-east-1
   MINIO_USE_SSL=false
   MINIO_PUBLIC_URL=http://localhost:9000
   MINIO_ROOT_USER=opensolve          # only used by docker compose
   MINIO_ROOT_PASSWORD=opensolve-secret

2. Start MinIO (or run the full stack with docker compose up --build):

   docker compose up -d minio minio-mc

   • API: http://localhost:9000
   • Console UI: http://localhost:9001

3. Provision the bucket (one-time). The bundled MinIO Client helper is the minio-mc service:

   docker compose up minio-mc

   The helper exits after calling mc mb --ignore-existing local/$MINIO_BUCKET, so it’s safe to rerun.

4. Run the app (docker compose up web). Upload routes will now stream directly into your MinIO bucket. In production, point the same variables at any S3-compatible endpoint (e.g., AWS S3, DigitalOcean Spaces, Cloudflare R2) and update MINIO_PUBLIC_URL to whatever domain/CDN exposes the objects.

If the storage variables are omitted the upload endpoints throw a descriptive error, so you can disable attachments entirely if desired.

🧑‍⚖️ Judge Worker & RabbitMQ

Phase 7 introduces a standalone judge worker that consumes RabbitMQ queues and runs submissions inside Docker sandboxes.

1. Configure env vars

   JUDGE_RABBIT_URL=amqp://opensolve:opensolve@localhost:5672
   JUDGE_RABBIT_PREFETCH=2
   JUDGE_SANDBOX_DRIVER=docker   # or mock for simulator mode
   JUDGE_SANDBOX_WORKDIR=/tmp/opensolve-judge

2. Start RabbitMQ + worker

   • Dev mode: docker compose up rabbitmq worker
   • Bare metal: run npm run judge:worker alongside npm run dev

3. Manual review tools — staff can review hybrid/manual submissions at /staff/judge/manual, posting MANUAL_ACCEPTED, MANUAL_PARTIAL, or MANUAL_REJECTED verdicts with notes/score.

When RabbitMQ is unavailable (or the worker is down) the app falls back to the inline simulator so basic flows keep working.

---

✅ Testing

Integration tests exercise Prisma CRUD flows against a disposable Postgres schema.

1. Ensure Postgres is running and migrations have been generated (npx prisma migrate deploy).
2. Provide the required environment variables (the tests read from .env; at minimum set DATABASE_URL, SESSION_SECRET, and APP_URL).
3. Run:

npm run test

Vitest will clone the schema into test_<worker> automatically, apply migrations, and run the CRUD suites without touching your development data. Use npm run test:watch during development for faster feedback.

---

📈 Metrics & Observability

• scrape GET /api/internal/metrics with Authorization: Bearer $METRICS_ACCESS_TOKEN.
• Exposes opensolve_trpc_duration_seconds, opensolve_submission_events_total, opensolve_judge_queue_messages, and contest gauges.
• Deployments should call this endpoint after rolling out to ensure judge queues and tRPC calls are healthy.

---

🛡️ Creating an admin account

Roles gate problem authoring, publishing, and moderation tools. To promote one of your users to ADMIN:

1. Seed or sign up the user as usual so they exist in the User table.
2. Run Prisma Studio (or any SQL client) and update the role column to ADMIN.

npx prisma studio

Open the User table, locate the account, and change the role dropdown to ADMIN. The change takes effect immediately—sign back in and you’ll see the staff console plus every curator tool. For scripted environments you can run:

npx prisma db execute --script "UPDATE \"User\" SET role = 'ADMIN' WHERE email = 'you@example.com';"

Remember: admins can publish/archive problems, edit roles, and bypass reviewer restrictions, so keep these accounts limited.

---

🛠 Deployment & Ops

• Environment parity: ops/env/*.env hold canonical dev/staging/prod configs. Root docker-compose.yml is the contributor default; staging and production compose specs live under ops/docker/.
• CI/CD pipeline: .github/workflows/ci-cd.yml runs lint → tests → Playwright → Docker builds → image pushes → staged releases.
• Deploy targets: use the docker-compose stacks under ops/docker/ for dev/staging/VM setups or the fly.toml spec for Fly.io. Secrets live in ops/env/*.env.
• Runbooks: outages, judge issues, DB slowdowns, and restore drills live under docs/quality.