December 2025 Trending GitHub Repositories in AI and Automation

This December, the GitHub trending list highlights several innovative projects focusing on AI-driven automation, multi-agent coordination, AI-powered development tools, and visual intelligence. The following sections categorize these repositories by their core functionalities and application domains.

---

AI-Powered Mobile and Desktop Intelligent Assistants

Open-AutoGLM

• Description: A mobile intelligent assistant framework based on AutoGLM. Controls devices via ADB/HDC, integrates vision-language models to understand screens, and automatically executes operations. Supports mainstream Chinese apps, remote debugging, sensitive operation confirmations, and manual takeover mechanisms. Models can be deployed locally or through third-party APIs, with multilingual and customized configurations.
• Stars: 19,396
• Created At: December 8, 2025
• Link: Open-AutoGLM

Roubao

• Description: The first open-source AI phone automation assistant requiring no PC. Native Android app in Kotlin using vision-language models. Supports Shizuku permission management without Root or ADB connections, completing screenshots, analysis, and device operations locally. Features a two-layer agent architecture (Tools/Skills), delegation via DeepLink, multi-model support (Qwen, GPT-4V), AES encrypted API key storage, and fully bilingual UI with Material 3 design.
• Stars: 737
• Created At: December 9, 2025
• Link: Roubao

---

Multi-Agent AI Development Frameworks and Agent Orchestration

Auto-Claude

• Description: Autonomous multi-agent coding framework where AI agents plan, build, and validate software using Claude AI. Features include parallel execution, isolated git workspaces, self-validating QA, AI-assisted merges, and persistent memory for cross-session learning. Available as a cross-platform desktop app with CLI support under AGPL-3.0 license.
• Stars: 3,435
• Created At: December 4, 2025
• Link: Auto-Claude

Oh My OpenCode

• Description: A powerful agent harness orchestrating multiple AI models (Claude, GPT, Gemini) as specialized teammates for coding tasks. Supports background agents, LSP/AST tooling, multimodal functions, and Claude Code compatibility. Includes smart context management, automated workflows, and production-grade optimizations. Offers extensive customization and works out of the box.
• Stars: 2,958
• Created At: December 3, 2025
• Link: Oh My OpenCode

Skills by GuDaStudio

• Description: Modular Agent Skills collection enabling seamless collaboration among Claude, Codex, Gemini, and others. Provides one-click installation and multi-platform compatibility. Contains strict workflows including context retrieval, multi-model analysis, prototype generation, code refactoring and audit. Only generates diff patches under Claude’s control to ensure security and code quality, with no file write permissions.
• Stars: 711
• Created At: December 16, 2025
• Link: Skills

---

AI-Assisted Browser Automation and Proxy Management

Vibium

• Description: A browser automation tool for AI agents featuring a Go-based binary (“Clicker”) managing Chrome via WebDriver BiDi, exposing an MCP server over stdio. Offers zero-setup browser control for LLMs like Claude Code with JS/TS clients supporting synchronous and asynchronous APIs. Automatically installs Chrome for Testing, manages cross-platform browser controls, and provides auto-wait, screenshots, and element interaction.
• Stars: 1,105
• Created At: December 12, 2025
• Link: Vibium

ProxyCast

• Description: A local desktop proxy that converts OAuth credentials from multiple AI clients (Kiro, Gemini, Tongyi Qianwen) into standardized OpenAI API tokens. Supports cross-platform unified management, automatic credential refresh, quota overflow switching, model fallback, and multi-routing. Compatible with OpenAI and Anthropic APIs. Data processing is local with HTTPS reverse proxy and remote API management for efficient reuse of AI quotas.
• Stars: 745
• Created At: December 13, 2025
• Link: ProxyCast

---

AI-Enhanced Visual Tools and Prompt Engineering

Prompt Fill

• Description: A structured prompt generation tool specifically for AI art, allowing conversion of text templates with {{variable}} syntax into visual fill-in forms. Features dynamic vocabulary systems with category management, customizable extensions, bulk import, drag-and-drop insertion, multi-instance independent selection, and real-time preview. Data is stored locally in-browser with options for HD image export and JSON backup. Offers template clone, cover upload, and color coding for enhanced user experience.
• Stars: 1,280
• Created At: December 6, 2025
• Link: Prompt Fill

PPT Master

• Description: AI-driven multi-role collaboration system converting documents into high-quality SVG visual content suitable for PPT, social media, and other formats. Implements Strategist, Executor, and Optimizer roles following CRAP design principles to output pure vector graphics with PPTX export. Includes chart templates, icon libraries, and post-processing tools to automate content decomposition and visual optimization workflows.
• Stars: 643
• Created At: December 10, 2025
• Link: PPT Master

---

AI Privacy and Image Processing Tools

Gemini Watermark Remover

• Description: A 100% client-side tool implemented in JavaScript to remove Gemini AI watermarks with mathematically precise Reverse Alpha Blending for pixel restoration. Automatically detects watermark sizes (48×48 or 96×96) and applies appropriate alpha maps for lossless removal. Runs fully in-browser with no image uploads, preserving user privacy. Includes a userscript to remove watermarks directly on Gemini conversation pages.
• Stars: 1,149
• Created At: December 19, 2025
• Link: Gemini Watermark Remover

---

This diverse collection of projects reflects rapid advancements in AI-enabled automation, collaborative coding, visual AI tools, and secure interaction layers, demonstrating robust innovation for developers, enterprises, and end-users.