MCP Catalogs
Homeosaurus screenshot

osaurus

by osaurus-ai·5,287·Score 59

A native macOS AI harness with MCP server support for local/cloud models, agents, and tools.

ai-llmdeveloper-toolsproductivity
277
Forks
61
Open issues
this month
Last commit
2d ago
Indexed

Overview

Osaurus is a comprehensive AI framework built in Swift for macOS, designed to provide a personal AI experience with persistent memory, autonomous agents, and cryptographic identity. As an MCP server, it exposes tools through a stdio bridge, enabling compatibility with MCP clients. The application supports various AI models including local MLX-optimized models, Apple's Foundation Models, and cloud providers like OpenAI and Anthropic. Its sandboxed environment allows safe code execution in isolated Linux VMs, making it suitable for both personal and professional use cases requiring security and privacy.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Personal AI assistant with persistent memory across sessions
you:Development workflow automation with code execution in sandboxed environment
you:Integration of MCP tools from various providers (GitHub, Notion, Vercel, etc.) into AI workflows
you:What models does Osaurus support?
you:How does Osaurus handle security and privacy?

When to choose this

Choose Osaurus if you want a native macOS AI agent with persistent memory, sandboxed execution, and MCP server capabilities that works both offline and with cloud providers.

When NOT to choose this

Don't choose Osaurus if you need cross-platform support or are looking for an open-source alternative with more flexible licensing options.

Tools this server exposes

12 tools extracted from the README
  • query_postgres

    Query PostgreSQL database

  • create_github_issue

    Create a new GitHub issue

  • list_files

    List files in a directory

  • search_web

    Search the web for information

  • send_email

    Send an email

  • schedule_task

    Schedule a task to run later

  • read_calendar

    Read calendar events

  • execute_shell

    Execute shell commands in a sandboxed environment

  • analyze_image

    Analyze images using vision capabilities

  • save_memory

    Save important facts to long-term memory

  • create_agent

    Create a new AI agent with specific capabilities

  • run_agent

    Run an agent with a specific task

Note: Tool names inferred from the plugin system description and MCP server documentation. The README mentions plugins for Mail, Calendar, Vision, macOS, Git, Filesystem, Search, etc. but doesn't provide specific tool names. Some tools were extra

Comparable tools

lmstudio-mcpnativeaioobabooga-text-generation-webui

Installation

Install via Homebrew:

brew install --cask osaurus

Or download the latest .dmg from [Releases](https://github.com/osaurus-ai/osaurus/releases/latest).

For Claude Desktop, add to claude_desktop_config.json:

{
  "mcpServers": {
    "osaurus": {
      "command": "osaurus",
      "args": ["mcp"]
    }
  }
}

FAQ

What models does Osaurus support?
Osaurus supports MLX-optimized local models, Apple's Foundation Models (on macOS 26+), Liquid AI LFM models, and cloud providers like OpenAI, Anthropic, Gemini, xAI/Grok, Venice AI, OpenRouter, Ollama, and LM Studio.
How does Osaurus handle security and privacy?
Osaurus uses cryptographic identities with secp256k1 keys, sandboxed code execution in isolated Linux VMs, and can run fully offline with local models. Cloud providers are only contacted when explicitly chosen by the user.

Compare osaurus with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.