GKD is an open-source Android application that automates UI interactions, primarily to skip ads, auto-confirm dialogs, and perform custom click actions. It operates as an accessibility service, using a custom selector language (similar to CSS) to match UI nodes and execute actions defined in remote subscription rules. The app is built with Kotlin and Jetpack Compose, leveraging Room for local persistence, Ktor for HTTP, and Coroutines/Flow for reactive state management. It also integrates with Shizuku for privileged operations like input injection and screenshot capture. The architecture is modular: the main `gkd-app` module contains the UI, services, and rule engine; `gkd-selector` is a Kotlin Multiplatform library for parsing and matching the custom selector syntax; `gkd-db` provides a multiplatform database layer; and `gkd-hidden-api` contains stubs for Android hidden APIs used for automation. Key entry points include the accessibility service (`A11yService`), the subscription repository (`SubscriptionRepository`), and the snapshot capture system. Data flows involve downloading subscription rules, parsing them into internal representations, matching against accessibility node trees, and executing actions (click, swipe, etc.). External dependencies include AndroidX libraries, Ktor, Kotlin Serialization, Coil, and the Shizuku privilege library. The codebase demonstrates strong engineering practices with extensive unit tests, a well-defined selector engine, and reusable patterns like `Loadable` for async state and `MutexState` for concurrency. However, it relies heavily on hidden Android APIs and a custom DSL, which introduces maintenance risk. Overall, it is a high-quality, actively maintained project with significant reuse potential in its selector engine and utility components.
django-click is a library that bridges the Click CLI framework with Django's management command system. It allows developers to write Django management commands using Click's decorator-based API instead of the traditional Django BaseCommand class. The library provides a drop-in replacement for Click's `command` and `group` decorators that automatically register the decorated function as a Django management command, handling all the integration details such as argument parsing, help output, verbosity, settings, and color support. The architecture consists of adapter classes (`DjangoCommandMixin`, `CommandAdapter`, `GroupAdapter`) that inherit from Click's `Command` and `Group` classes and override key methods (`run_from_argv`, `create_parser`, `execute`, `print_help`) to conform to Django's command execution protocol. A `BaseRegistrator` class handles the registration process: it extracts the command name from the module name, applies Click decorators with common Django options (verbosity, settings, pythonpath, traceback, color), and sets the module-level `Command` attribute to a lambda that returns the Click command instance. This enables seamless integration with both `manage.py` invocation and `call_command()` usage. The library also provides a `pass_verbosity` decorator for accessing the Django verbosity level within Click commands, and a `ModelInstance` custom parameter type that allows commands to accept Django model instances as arguments with support for custom lookups. The code is well-tested with comprehensive test coverage for CLI invocation, `call_command` integration, error handling, color support, and edge cases like abort handling and stdout/stderr capture.
Memo is a Python CLI tool for managing Apple Notes and Apple Reminders directly from the terminal. It uses Click for command-line interface and AppleScript (via osascript) to interact with macOS's Notes and Reminders applications. The application provides commands to list, view, add, edit, delete, move, search, and export notes, as well as manage reminders (list, complete, add, delete, edit). Notes are fetched as HTML from Apple Notes, converted to Markdown for display and editing, and images are handled via a placeholder mechanism to preserve inline images through edit cycles. The architecture is a standard CLI application with helper modules for each operation, caching for notes, and validation for flag combinations. It is well-tested with pytest, uses modern Python 3.13, and has a professional CI/CD pipeline and documentation site.