MCP.so
登录

MCP-NixOS - Because Your AI Assistant Shouldn't Hallucinate About Packages

@utensils

关于 MCP-NixOS - Because Your AI Assistant Shouldn't Hallucinate About Packages

MCP-NixOS - Model Context Protocol Server for NixOS resources

基本信息

分类

AI 与智能体

许可证

MIT

运行时

python

传输方式

stdio

发布者

utensils

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "nixos": {
      "command": "uvx",
      "args": [
        "mcp-nixos"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is MCP-NixOS?

MCP-NixOS is an MCP server that provides accurate, real-time information about NixOS packages, options, Home Manager, nix-darwin, Nixvim, FlakeHub, Noogle, NixOS Wiki, nix.dev, package versions (via NixHub), binary cache status, and local flake inputs. It integrates with multiple APIs and is designed for AI assistants that need reliable NixOS data without hallucinations. It works on any system (Windows, macOS, Linux) and does not require Nix installed.

How to use MCP-NixOS?

Install and run via uvx mcp-nixos (recommended), pip install mcp-nixos, nix run github:utensils/mcp-nixos, or Docker (ghcr.io/utensils/mcp-nixos). Configure as an MCP server in your client (e.g., Cursor, Claude, Pi) using the nixos server name. Two tools are provided: nix (unified query with actions like search, info, stats, cache, flake-inputs) and nix_versions (package version history). Transport defaults to STDIO; set environment variables MCP_NIXOS_TRANSPORT=http and optionally MCP_NIXOS_HOST, MCP_NIXOS_PORT, MCP_NIXOS_PATH, MCP_NIXOS_STATELESS_HTTP for HTTP mode.

Key features of MCP-NixOS

  • Two consolidated tools (~1030 tokens total).
  • No Nix installation required (runs anywhere Python runs).
  • Queries 130K+ packages and 23K+ NixOS options.
  • Supports STDIO and HTTP transports.
  • Includes binary cache status checking.
  • Accesses FlakeHub, Noogle, NixOS Wiki, nix.dev, and NixHub.

Use cases of MCP-NixOS

  • Search for NixOS packages or options and get real-time, accurate results.
  • Check binary cache

评论

AI 与智能体 分类下的更多 MCP 服务器