Installation

Choose the path that fits your machine on Download & run:

The newest stable desktop build is always available from the latest GitHub release. The website buttons below follow that same published release.

• Browser UI — source mode on macOS, Windows, or Linux.
• macOS app — Apple Silicon DMG, with no Python or terminal required.
• Windows app — x64 EXE or MSI, with no Python or terminal required.

The website includes platform-aware buttons that link directly to the matching file in the newest stable GitHub release.

Unsigned installers — first install only

Desktop installers are working but not OS-signed: macOS Gatekeeper and Windows SmartScreen warn on first install because there is no paid code-signing certificate yet (see Signing & notarization). Updates are different: they are cryptographically signed and verified in-app, and don’t re-trigger those warnings. Verify manual downloads against SHA256SUMS.txt on the release.

Desktop app (recommended)

Use the direct buttons on the Golavo website, or open the latest release:

• macOS (Apple Silicon) — the .dmg. Open it and drag Golavo into the Applications folder, then launch it from Applications. This step matters: an app run straight from the disk image sits on a read-only volume and cannot update itself (the app will remind you if you forget).
• Windows (x64) — the -setup.exe (NSIS). The .msi also works; both update in-app afterwards.

On first launch the OS will warn (unsigned build):

• macOS — “Golavo can’t be opened because Apple cannot check it for malicious software.” Right-click the app → Open → Open, or clear the quarantine flag: xattr -dr com.apple.quarantine /Applications/Golavo.app.
• Windows — SmartScreen shows “Windows protected your PC.” Click More info → Run anyway.

Optional: add local AI with Ollama

Golavo’s deterministic match analysis works without AI. If you want the optional Fast or Deep narrative, open Settings → Local intelligence after installation:

1. Use Golavo’s link to install Ollama for macOS, then keep Ollama open.
2. Choose Check again so Golavo can confirm the loopback service is ready.
3. Download the recommended Fast or Deep model inside Golavo. The app shows progress, transferred size, and a cancel action, then assigns the model automatically.

The same guide appears beside the AI analysis controls. Model installation is explicit, stores the model locally through Ollama, and never uploads match data. Full details: AI providers & local models.

Staying up to date

From v0.2.1 on, Golavo updates in-app: a one-time card asks whether to check for updates automatically (your choice; a manual Check now always lives in Settings → Updates), and installs happen only when you click. Every update is signature-verified before install and your ledger is backed up first — see Updates & rollback.

When an automatic or manual check finds a newer stable GitHub release, Golavo shows a visible in-app notification and keeps an Update available button in the header until you review, install, or skip that version. The macOS and Windows installers linked from the website are release builds with this updater compiled in; the stable release workflow refuses to publish them if updater signing is unavailable.

Coming from v0.1.0 or v0.2.0? Those builds predate the updater — download and install the current version manually once; it’s in-app from then on.

Uninstalling on Windows

The uninstaller offers to delete application data. Leave that unchecked if you want to keep your prediction ledger — it lives in that data folder.

Browser UI from source

This is a local web app: the API listens only on 127.0.0.1, the UI opens at http://127.0.0.1:5173, and nothing is hosted by Golavo.

git clone https://github.com/udhawan97/Golavo.git
cd Golavo
cp .env.example .env     # optional; Golavo runs local with no keys
make setup               # install core + server + ui dev deps
make dev                 # start both services and open the browser UI

Press Ctrl+C in that terminal to stop both local services. To start without opening a new browser tab, use python scripts/dev.py --no-open.

Source checkouts update with git pull; the in-app updater stays out of the way (Settings says so instead of showing dead controls). See Build from source for prerequisites.

Build the desktop app locally

Prerequisites: Rust (stable), Node 22+, Python 3.12+, and the Tauri system dependencies for your OS. Then:

# from the repo root
packaging/build.sh aarch64-apple-darwin      # or x86_64-pc-windows-msvc, etc.

The bundle and a per-target SHA256SUMS-<target> land in packaging/out/:

• macOS — a .dmg, plus the .app.tar.gz updater artifact when the updater-signing key is present. The unpacked .app remains under Tauri’s target bundle tree.
• Windows — an .msi (WiX) and an -setup.exe (NSIS) installer.

CI (.github/workflows/release.yml) builds these on native macOS and Windows runners for every v* tag, assembles the update manifest, and publishes the release atomically (draft until every asset is uploaded).

Signing & notarization

• macOS notarization requires the Apple Developer Program ($99/yr) for a Developer ID certificate. Without it, no notarized release is possible.
• Windows signing uses SignPath Foundation (free for OSS) or Azure Trusted Signing.
• Signed auto-update is active: updates are signed with the Tauri updater key in CI and verified in-app before installing. This is independent of (and unaffected by) the OS-level signing above.

The first two are still gated on secrets the project does not hold; the pipeline skips those steps honestly rather than faking them. See Updates & rollback.