24 resources
Jev Ultrafast is an extremely fast Python-based web agent that enables efficient browser automation through dynamically indexed action spaces. It uses TypeSafe LLM to perform clicks, inputs and other operations, significantly reducing latency and costs, and is suitable for scenarios where web page interaction tasks need to be completed quickly.
Zuse is an open source desktop application for developers, integrating AI programming agents such as Claude and Codex. Provides a persistent project-aware interface, including structured chat, file viewer, terminal and other functions. Supports macOS and x64 Linux platforms, and requires local installation of the corresponding agent CLI.
Cida is a native Mac application that uses large models selected by the user to translate and polish text at any location. Supports real-time translation of selected text, screenshot recognition translation and text polishing functions. Focusing on privacy, the API Key is stored in the keychain, no data is collected, and it is compatible with multiple AI service provider interfaces.
xcertplay is an Android car CarPlay receiver that supports wired and wireless connections. iPhone pairing is achieved through the MFi chip or CH341 bridge, and it is compatible with Android and Android Automotive OS. It provides multi-channel audio output and dynamic resolution adaptation for voice, navigation, etc., aiming to enable Android cars to have CarPlay functions.
Laya-MLX is a native Python library running on Apple Silicon for performing ultra-fast (7-14ms) typed decision-making tasks. It is based on the MLX framework, does not require PyTorch or cloud API, supports offline local reasoning, and is suitable for scenarios that require low-latency structured output.
Rizzo Flow is an open source, local-first LLM decision-making tool compatible with the Jev API. It transforms unstructured state input into typed probabilistic decisions without generating text tokens. Supports the llama.cpp backend and can run in Metal, CUDA, Vulkan, ROCm, SYCL and CPU environments. It is suitable for local AI application development that requires low latency and high privacy.
High-performance code-intelligent MCP server that indexes the code base into a persistent knowledge graph. Supports 158 languages, millisecond-level indexing and sub-millisecond queries, significantly reducing token consumption. Provides a single static binary file with zero dependencies, specially designed for AI programming agents to improve code understanding efficiency.
ZCode is an AI programming workbench that provides desktop applications, browser interfaces and terminal Agents. This warehouse contains client, back-end services, shared UI, Agent CLI and runtime source code, supports multi-entry development, and aims to create a powerful, intelligent and scalable programming environment.
Status Trio is a native macOS app that combines Wi-Fi, battery, and volume status into a single compact icon. It supports display in the menu bar or Dock, providing a highly configurable status display inspired by the iPhone Duo combined status bar design and designed to simplify system status monitoring.
OnCo is a public oncology knowledge graph that aims to provide comprehensive information on cancer, treatment and research. It includes website, JSON API, MCP server and CLI tools to support patients, doctors and AI agents to obtain structured medical data and correlation analysis.
This is an experimental controller that allows TypeSafe's Jev model to directly select the control input of NES "Super Mario Bros." through structured JSON instead of screenshots. It parses simulator telemetry data, automatically performs game operations, and is used to explore AI's decision-making capabilities in structured states.
SleekFin is a custom theme plug-in for Jellyfin that provides modern interface redrawing. The main features include dark skin, configurable homepage hero area, floating navigation bar and optimized details page layout, aiming to improve the visual experience and personalization capabilities.
DocCloak is a browser-based document anonymization tool that runs locally using WebAssembly. It can automatically detect and replace sensitive personal information (PII) in documents to prevent data leakage to AI services, while supporting the restoration of placeholders in AI responses to their original names. No server required, ensuring privacy and security.
M3E Canvas is a browser-based prototyping tool that supports the use of Material 3 Expressive specifications to draw interactive interfaces. Users can link pages, simulate clicks, and generate prompt words for use by AI programming tools with one click. No backend is required, and the data is stored locally.
Doska is a local-first, lightweight Kanban application that supports Markdown editing. Data is first stored in IndexedDB and can be synchronized to folders or servers to achieve cross-device collaboration. It has offline availability, multi-user real-time synchronization, SSO and other functions, aiming to provide a fast, minimalist and feature-rich task management experience.
Vibe Squad is a Python-based multi-model AI orchestration tool that defines behaviors through Markdown. It uses a coordinator to distribute tasks to 71 professional roles, execute them in parallel in isolated git workspaces, and supports independent reviews across model families (such as Claude, Codex, etc.), aiming to provide an efficient and safe automated workflow for structured coding.
masume is a terminal database client that supports browsing object trees, generating ER diagrams, SQL queries and result grid editing. It has built-in syntax highlighting, auto-completion and query plan visualization functions, and integrates Notebook mode. The database directory can be shared with the AI agent to improve development efficiency.
Infomarchy is a plug-in for Omarchy desktop that converts wallpapers into real-time information panels. It monitors and displays the status, cost, rate limit and machine performance of all running AI agents (such as Claude, Ollama), and supports 7-day heat maps and recent prompt words. The display can be quickly switched through shortcut keys, helping users understand system resources and AI usage at a glance.
Kitter is a lightweight Agent Skills manager developed based on Rust, providing desktop applications and CLI. It allows users to flexibly configure the required skill combinations for different projects through centralized management of skill libraries, and supports one-click updates, effectively solving the problem of confusing skill versions in multiple projects.
This warehouse is an AI-driven PRD generation tool that clarifies demand gaps through interactive questionnaires and transforms messy materials into clearly structured and acceptable product requirements documents. It supports the creation and continuous updating of master PRDs to ensure one-to-one correspondence between business requirements and acceptance standards, reducing communication costs and improving R&D and testing efficiency.
AppStore Forge is a native Mac application used to convert original application screenshots into store materials that comply with App Store and Google Play specifications. No accounts, uploads or servers required, data remains entirely local. Supports customization of device borders, backgrounds, titles and panorama layouts, and provides WYSIWYG preview and export functions.
Octoscode is the terminal interface of Octos AI coding assistant, similar to Claude Code. It provides a keyboard-driven TUI that supports chat, code difference viewing, tool approval, and background tasks, allowing you to efficiently interact with AI without leaving the shell. After installation, the server is automatically configured to support local AI provider integration, improving developers' coding experience in the command line.
Covan is a team-oriented shared AI agent platform that supports self-hosting. Team members can jointly train an AI agent with a unified knowledge base while enjoying private session isolation based on Postgres row-level security. It has knowledge package management, pgvector-based citation answers, automated routine tasks and collaboration functions, aiming to provide intelligent assistants for the entire company, not just engineers.
OpenClaw-RPA is an AI RPA compiler that records browser, Office and API operations and generates independent Python script playback. No LLM calls required, zero token cost, fast and no illusions. Supports session persistence and cross-platform operation (including Docker), suitable for automating repetitive tasks.