AstrBot is a production-grade, multi-platform LLM chatbot framework and development platform. It connects to 10+ instant messaging platforms (QQ, Telegram, Discord, Slack, WeChat, Lark, etc.) via adapter pattern, integrates with 20+ LLM providers (OpenAI, Anthropic, Gemini, DashScope, etc.), and provides a plugin system (Stars) with command/regex/filter decorators. The architecture is a modular monolith: a FastAPI backend serves a Vue 3 dashboard for management, while a pipeline of stages processes incoming messages through waking checks, rate limiting, content safety, LLM interaction, and response delivery. It includes a knowledge base with RAG (FAISS vector store, chunking, hybrid retrieval), agent runners (Dify, Coze, DeerFlow), sandboxed code execution (computer use), TTS/STT, cron jobs, and a CLI. The codebase is well-structured with extensive test coverage, but its broad scope introduces some duplication and deprecated modules.
Lux is a fast and simple video downloader CLI tool written in Go. It supports downloading videos, images, and audio from over 40 websites including YouTube, Bilibili, TikTok, Instagram, Twitter, and many Chinese streaming platforms. The tool operates as a single binary with a rich set of command-line options for selecting quality, downloading playlists, resuming interrupted downloads, multi-threaded downloading, and integration with aria2 for RPC-based downloads. Architecturally, Lux follows a plugin-based extractor pattern. Each supported website has its own extractor module under `extractors/`, registered via `init()` functions into a global map keyed by domain. The main entry point is `main.go`, which parses CLI flags via `urfave/cli` and dispatches to the appropriate extractor. Extracted video metadata (streams, quality, size) is passed to the downloader, which handles HTTP requests, progress bars, file merging via ffmpeg, and optional subtitle embedding. Key data flows: URL input → domain detection → extractor selection → HTML/API scraping → stream metadata extraction → downloader (single or multi-threaded) → file output. External dependencies include ffmpeg for merging and subtitle embedding, and various third-party Go libraries for HTTP requests, HTML parsing, JavaScript execution (for TikTok signing), and progress bars. The codebase is well-organized with clear separation of concerns, extensive test coverage for individual extractors, and CI workflows that run weekly to verify extractor functionality. While many extractors rely on fragile regex-based scraping of HTML pages, the overall architecture is solid and the downloader implements robust features like resume support and configurable retry logic.
ONO is an Android Xposed module that modifies and enhances the QQ messaging application (Tencent). It provides a broad set of features including custom sticker panels, message encryption, anti-revoke, developer tools (packet inspection, cookie retrieval), UI tweaks (bubble redirect, profile card menu), and entertainment hooks (fake group owner, sport step spoofing). The module supports both legacy Xposed (API 51–99) and modern LSPosed (libxposed API 100) frameworks, with a dual-entry architecture that unifies startup through a shared `UnifiedEntryPoint` and `StartupRoutine`. Internally, it uses a sophisticated hook item framework based on annotations and a KSP annotation processor to automatically register hook classes. The codebase is heavily reliant on reflection and reverse-engineering of QQ's internal classes, making it tightly coupled to specific QQ versions, but the core utilities (reflection helpers, configuration management, protocol building) are well-crafted and potentially reusable in other Android hooking projects. The project is actively maintained with frequent commits, responsive issue triage, and a comprehensive build system using Gradle convention plugins.
The RevokeMsgPatcher is a Windows Forms application designed to patch instant messaging clients (primarily Tencent's WeChat and QQ) to prevent message recall and enable multiple simultaneous instances. The application operates by modifying binary files of the target applications at specific offsets or through pattern-based searches, guided by a version-specific patching database stored in JSON format. The architecture is monolithic, consisting of a single executable that embeds patching data as resources. Key components include a user interface for selecting the target application and version, a patching engine that applies the binary modifications, and a data layer (exemplified by JsonData.cs) responsible for parsing the JSON patch definitions. The workflow involves user selection, loading of the appropriate patch set, application of patches to the target binaries, and launching the patched application. While the tool effectively achieves its goals, it faces inherent challenges due to the fragile nature of binary patching. The patching logic is tightly coupled to specific byte sequences, requiring frequent updates to keep pace with target application releases. The project employs a two-tier strategy (exact offset patches for specific versions and pattern-based ranges for version intervals) to mitigate this, but maintenance remains high. Despite these challenges, the project demonstrates practical engineering in managing version-specific patches and provides a reusable pattern for JSON-driven binary modification systems.