Osaurus

•
8.2k
•

Powers AI workflows on macOS with an always-on edge runtime for local and cloud models, enabling shared tools via Model Context Protocol.

Osaurus is a comprehensive AI edge runtime specifically designed for macOS, leveraging Apple Silicon for optimized performance. It allows users to run both local large language models (LLMs) via MLX and connect to various cloud providers like Anthropic and OpenAI. Functioning as a Model Context Protocol (MCP) server, Osaurus facilitates the sharing of custom tools and capabilities across different AI applications. It offers a robust framework for building and automating AI workflows through custom agents, a multi-layer memory system, a plugin architecture, and features like schedules and watchers for autonomous task execution.

Key Features

01Optimized local LLM inference with MLX Runtime for Apple Silicon
02MCP Server and Remote MCP Providers for tool sharing and aggregation
03OpenAI, Anthropic, and Ollama API compatibility for local and remote models
04Autonomous Work Mode with issue tracking, parallel tasks, and file operations
05Customizable AI Agents with unique prompts, tools, themes, and memory
063,564 GitHub stars

Use Cases

01Automating recurring AI tasks like daily journaling or report generation with schedules and watchers
02Creating specialized AI assistants for coding, research, or creative writing with custom agents and skills
03Extending AI capabilities by integrating new tools and building a personal knowledge base through its memory system