Questarr

Questarr

A video game management application inspired by the -Arr apps (Sonarr, Radarr, Prowlarr…) and GamezServer. Track and organize your video game collection with automated discovery and download management.

⭐ Star us on GitHub — your support motivates us a lot! 🙏😊

Table of Contents

• Questarr
  • Table of Contents
  • List of features
  • Installation
  • Screenshots
  • Release classification
  • Tech Stack
  • Configuration
  • Roadmap
  • Troubleshooting
  • Project Security & Documentation
  • Contributing
  • License
  • Acknowledgments

List of features

Supported Indexers/Downloaders

• Sync indexers from Prowlarr.
• G4u.to: the site offers an API to VIP members, which can be used in Questarr.

Installation

Docker is the easiest way to deploy Questarr with all dependencies included. Since 1.5.0, native Windows installation is also supported via the Windows installer. Questarr uses SQLite by default, with optional PostgreSQL support. Persist the data volume to retain the SQLite database, configuration, and encryption keys.

Supported architectures: released Docker images are published for linux/amd64 and linux/arm64, so Questarr runs on a Raspberry Pi 4/5 with a 64-bit OS, other 64-bit ARM SBCs, and ARM-based NAS boxes as well as on x86 hardware. Docker selects the right architecture automatically — the commands below are identical on every platform. (32-bit ARM, e.g. armv7/a 32-bit OS on Raspberry Pi 3 and earlier, is not supported. The Home Assistant add-on is amd64-only.)

Option 1: One-liner (Simplest but minimal)

docker run -d -p 5000:5000 -v ./data:/app/data --name questarr ghcr.io/doezer/questarr:latest

Option 2: Docker Compose (more detailed)

1. Use the docker-compose.yml file from the repo or create a minimal one:

   services:
     app:
       image: ghcr.io/doezer/questarr:latest
       ports:
         - "5000:5000"
       volumes:
         - ./data:/app/data
       environment:
         - SQLITE_DB_PATH=/app/data/sqlite.db
       restart: unless-stopped
   

2. Start the application:

   docker compose up -d
   

3. Access the application: Open your browser to http://localhost:5000

4. Optional — use Postgres instead of SQLite:

   Questarr uses SQLite by default and needs no database configuration. If you would rather run it against Postgres, use the bundled compose file:

   POSTGRES_PASSWORD=$(openssl rand -hex 32) docker compose -f docker-compose.postgres.yml up -d
   

   Keep whatever password you pick — the same value must be supplied on every restart, and it must be safe to embed in a connection URL (a hex string from openssl rand -hex 32 always is).

   Switching an existing SQLite install to Postgres does not move your data — run scripts/sqlite-to-pg.ts to bring your library across. See docs/DATABASE.md.

Windows (installer)

Proxmox VE (LXC)

bash -c "$(curl --fail --silent --show-error --location --proto '=https' --proto-redir '=https' https://raw.githubusercontent.com/Doezer/Questarr/main/scripts/proxmox/questarr-lxc.sh)"

UNRAID

CasaOS

Umbrel

Cosmos Cloud

Kubernetes (Helm)

git clone https://github.com/Doezer/Questarr.git
cd Questarr
helm install questarr charts/questarr --namespace questarr --create-namespace

kubectl port-forward -n questarr svc/questarr 5000:5000

Home Assistant Add-on

All platform definitions live in the repository and share the same mounts and ports — see docs/HOME_SERVER_APPS.md for details.

Screenshots

Release classification

Smart release filtering, built in and free. Questarr recognizes patches, DLC, full games and non-game files (software, movies, soundtracks) right on your server, before anything is downloaded. No account, no API key, nothing leaves your network. Unsure cases can optionally be double-checked by a cloud model, and you stay in control: anything classified as a patch, DLC or non-game file is held for your review instead of being downloaded blindly (you can also hold releases the classifier cannot decide on).

Indexers return a lot of look-alikes: the game, a day-one patch, a DLC, a crack-only folder, a demo, a soundtrack. A fixed list of keywords cannot tell them apart reliably, because release names are inconsistent and the same words mean different things in different titles. Questarr classifies each release with a small machine-learning model and holds back the ones that are probably not the full game.

What you get

• Fewer wrong grabs. Before the automatic search sends a release to your download client, patches, DLC, expansions, crack-only releases, demos, soundtracks and other extras are held for your review instead of being downloaded.
• Works offline. The second layer is a local model (tf-idf with logistic regression) that runs on your machine. No account, no API key, and release names are not sent anywhere.
• Learns from names, not only keywords. It weighs the whole release name, so it handles spellings and group conventions that a hand-written rule list misses.
• Optional cloud model. In hybrid mode, the TypeSafe cloud model only receives the releases the rules and the local model could not settle (or all unsettled releases, if the local model cannot be loaded). In cloud mode it handles them without the local model. Use local mode to keep release names on your server.
• Gets better with your input. Optional active learning keeps the releases the classifier was unsure about so you can verify them in Settings and export the confirmed labels as CSV. Nothing is retrained automatically, a cloud suggestion is never stored as a verified label, and the labels and matches you provide stay on your server and are never sent to the cloud model. They are not sent anywhere else either, unless you tick Share my verified labels automatically in the review card (off by default, with a preview of the exact rows).

How it works

1. Fast rules catch the obvious cases.
2. The local model decides when it is confident enough (threshold 70% by default).
3. Only what is left can go to the cloud model, if you enabled it.

The local model is bundled with Questarr (including the Docker image), so it works out of the box. To use your own, place a release-classifier.json in <database folder>/models/ or point CLASSIFIER_MODEL_PATH at it; it takes priority over the bundled one (see The local model). If no model can be loaded, only the rules run locally, and hybrid sends every release the rules cannot settle to the cloud model, if one is configured.

Choose the mode in Settings → Release classification: hybrid (default), local (never leaves your machine) or cloud. The same card controls the confidence threshold, holding unclassified releases, and active learning. An environment variable, when set, wins over the saved setting and locks the field. The same pipeline types the manual search results and decides the automatic-download hold, so a release gets one type everywhere.

Details, environment variables and the dataset are in docs/RELEASE_CLASSIFICATION.md.

Tech Stack

• APIs: IGDB (game metadata), Torznab/Newznab (indexer search), PCGamingWiki, NexusMods, xREL.to
• AIs usage:
  • Claude and Github Copilot are used for AI-Assisted coding, internal code reviews, PR cleanup. Eventually automated coding and troubleshooting for small tasks and bug reports.
  • Gemini & Codex are used for automated code reviews, and brainstorming from time to time (as well as Perplexity for this usage).
  • Google Jules is used for light periodical refactoring.

Configuration

1. First-time setup:

• Create your admin account
• Configure the IGDB credentials

Once logged-in:

• Configure indexers
• Add downloaders
• Add games!

See Configuration on the Wiki for more detailed info.

git clone https://github.com/Doezer/Questarr.git
cd Questarr

docker-compose up -d

git pull
docker-compose down
docker-compose build --no-cache
docker-compose up -d

Roadmap

Based on the Product Requirements Document, here is the release status and remaining work, in priority order:

• ✅ P0 — Post-Processing Pipeline : Move/copy completed downloads to a destination path, extract archives — closing the “set it and forget it” loop.
• P1 — Smart Game Backlog: Installed-version tracking and update filtering are available; configurable automatic upgrades remain a priority.
• P2 — Direct Download Support: Add debrid services (Real-Debrid and similar) as a downloader option, no seeding required.
• P3 — External Library Sync: ✅ Manual owned Steam library import and local filesystem scans shipped; periodic owned-library reconciliation and GOG import remain planned.
• P4 — Integrations with External Tools: ✅ Playnite extension shipped (library sync, request-to-download); ✅ RomM shipped (platform folders, rescan, ownership sync); Gameyfin and a generic webhook for anything not explicitly supported are still planned.
• ✅ P5 — Indexer Page Links: A “View on indexer” link on search results and downloads.
• ✅ P6 — PostgreSQL Support: PostgreSQL is available as an optional backend, with SQLite remaining the zero-config default.

Ongoing: mobile responsiveness, search UX, performance, and security improvements.

See the full PRD for problem statements, detailed scope, and non-goals.

Troubleshooting

See Troubleshooting on the Wiki

If you run into an issue, go to the Logs page and click Send Logs before reporting it — it makes diagnosing the problem much easier.

Getting Help

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Discord: Join our Server

Project Security & Documentation

• Start with docs/GITHUB_DOCUMENTATION.md for the canonical documentation map.

Star History

Contributing

• See .github/CONTRIBUTING.md for guidelines on how to contribute to this project.
• See MAINTAINERS.md for the current list of project members with access to sensitive resources.

Contributors

Made with contrib.rocks.

Legal Disclaimer

Questarr is a self-hosted game manager designed solely for organizing, tracking, and automating game libraries using user-provided data and metadata APIs (such as IGDB). Questarr does not host, distribute, or provide any copyrighted game content, ROMs, or download links. It is a technology-neutral tool: any indexers, download clients, or sources you configure are chosen and operated entirely by you. You are solely responsible for ensuring that your use of Questarr, and any content you access or download through third-party services you configure, complies with all applicable laws and the terms of service of those third parties.

License

GPL-3.0-only - see COPYING for the full license text.

Acknowledgments

• Inspired by Sonarr and GamezServer
• Game metadata powered by IGDB API
• UI components from shadcn/ui
• Security scanning supported by Snyk through the Snyk Open Source program
• AI code reviews by CodeRabbit, free for open source
• This project is tested with BrowserStack