This daily update tracks GitHub repositories experiencing the highest relative star growth over a rolling seven-day window, prioritizing emerging momentum over established baseline popularity. By ranking by percentage gain rather than absolute counts, we surface high-velocity projects like Andee, which saw a 1,463% surge in interest by enabling on-device Android automation. This week’s leaders highlight a pivot toward localized, hardware-specific execution and specialized media production, exemplified by PS5LM’s ability to run GGUF models on jailbroken consoles and OpenFilm’s framework for converting coding agents into video editors.
| # | Repository | Stars Gained | Gain % | Overview |
|---|---|---|---|---|
| 1 | kuafuai/andee | 1946 | 1463.2% | An Android-based AI agent that operates mobile devices by interacting with the accessibility tree and screenshots to perform tasks like managing subscriptions or booking appointments. |
| 2 | AhmadHassan-BTed/B | 2351 | 697.6% | An agentic desktop companion that uses semantic screen awareness and LLM inference to proactively assist users during their workflow. |
| 3 | openfilm/openfilm | 1479 | 414.3% | An open-source AI video agent that enables coding agents to generate, edit, and render high-quality films using web technologies. |
| 4 | cobanov/PS5LM | 491 | 363.7% | A private, local LLM server that utilizes a jailbroken PS5’s GPU to serve language and decision models to other devices on a home network. |
| 5 | ling-stack-org/ling-stack | 736 | 275.7% | A local workspace manager that orchestrates multiple AI models and teammates into unified project channels for collaborative tasks. |
Andee (kuafuai/andee )
Andee is an autonomous AI agent designed to act as a resident intelligence on Android devices rather than a standard cloud-based assistant. It utilizes accessibility services to navigate apps, manage notifications, and execute complex routines called “scenes,” such as monitoring messages or automating repetitive tasks. The system prioritizes privacy by running the agent loop locally and storing personal memory in a device-only notebook. It is intended for users seeking a highly personalized, privacy-centric digital companion capable of interacting with any mobile application without requiring third-party SDK integrations.
B Desktop Companion (AhmadHassan-BTed/B )
B is an agentic desktop companion designed to live as a persistent, proactive overlay on a user’s workstation. It utilizes semantic screen sensing and OCR to understand active window content, allowing it to offer contextual insights or documentation without explicit user commands. The system features a dedicated ‘Work Mode’ for non-intrusive monitoring and supports both cloud-based and local LLM inference. It is intended for power users seeking an autonomous digital assistant that integrates seamlessly into their existing desktop environment.
OpenFilm (openfilm/openfilm )
OpenFilm is an open-source framework designed to bridge the gap between AI coding agents and professional video production. It allows users to prompt agents like Claude Code or Codex to generate films composed of HTML, CSS, and web-based graphics, which can then be refined via a local browser-based editor called Studio. Key features include timeline-based editing, support for various media providers like ElevenLabs and OpenAI, and the ability to export to professional formats like Final Cut Pro and DaVinci Resolve. This tool is ideal for developers and creators looking to automate video content creation through natural language instructions.
PS5LM (cobanov/PS5LM )
PS5LM transforms a jailbroken PlayStation 5 into a dedicated, private AI model server. By leveraging the console’s RDNA 2 GPU, it serves GGUF-formatted models via OpenAI and Anthropic-compatible APIs to laptops, phones, and coding tools. The system supports standard text generation and specialized decision models that provide typed responses with calibrated probabilities. It is designed for privacy-conscious users who want to run high-performance local inference without relying on cloud services or a dedicated PC.
Ling Stack (ling-stack-org/ling-stack )
Ling Stack is a local-first workspace designed to centralize interactions with various AI models, including OpenAI, Anthropic, and DeepSeek. It enables users to manage AI teammates within project channels, providing explicit controls for tool approvals, model bindings, and conversation history. Developers and power users can utilize the platform to coordinate complex workflows, such as document reviews or media production, while keeping all session data and artifacts stored locally. The system supports a wide range of providers through a unified Model Workbench.