Gear Lever is a GNOME desktop application for managing AppImage files on Linux. Its core function is integrating AppImages into the desktop environment by extracting them, generating .desktop entries with metadata and icons, and keeping them updated. It supports one-click install from downloaded AppImages, drag-and-drop integration, in-place version updates from multiple sources (GitHub releases, GitLab, Codeberg, Forgejo, FTP, and static URLs), side-by-side version retention, per-app environment variables and exec arguments, and bulk update operations. The application is distributed primarily as a Flatpak on Flathub and also ships a headless CLI (src/Cli.py) matching most GUI operations such as integrate, update, remove, list, and fetch-updates. Architecturally, this is an event-driven GTK4/Adwaita desktop application written in Python. The main window (GearleverWindow) uses Adw.Leaflet navigation with a drag-and-drop overlay, and views communicate through custom GObject signals. The core lifecycle logic lives in AppImageProvider, which handles download, extraction, desktop-entry generation, and launch. Extraction uses a fallback chain (dwarfs → 7zz → unsquashfs → appimage-extract) to handle the fragmented AppImage format landscape. Update sources are abstracted behind an UpdateManager ABC with separate backend classes per hosting platform. Configuration is split between a global JSON settings store and per-app INI files, with a migration path from an older JSON schema. Background work is managed through custom async/idle/debounce decorators that marshal tasks onto worker threads and safely return results to the GTK main loop. The system relies on external runtime binaries (7zz, dwarfsck, dwarfsextract, unsquashfs, readelf, strings, file, get_appimage_offset) in addition to Python packages (dbus-python, pyxdg, ftputil, requests, desktop-entry-lib). Notably, the repo embeds a demo.AppImage binary used as a test fixture, illustrating the extraction pipeline against a real-world AppImage. Test coverage is thin — only CLI integration tests that download actual AppImages — and the codebase shows some legacy baggage (ftputil, direct D-Bus usage, fragile external-tool version assumptions). Nonetheless, the code is modular, actively maintained with nightly builds, well-internationalized (16 languages), and demonstrates real craftsmanship in its abstraction layers.