AgentRunKit

Overview

AgentRunKit is a Swift 6 SDK for building LLM-powered agents with type-safe tool calling. It offers zero-dependency core, full Sendable support, async/await functionality, and works with both cloud and on-device inference via MLX on Apple Silicon. The SDK includes a robust MCP client implementation supporting stdio transport, tool discovery, and JSON-RPC communication.

Try asking AI

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

• get_weatherTool<WeatherParams, String, EmptyContext>

  Get the current weather for a city

  Try:What's the weather in San Francisco?Try:Check the weather for New York

Note: Inferred from code example showing tool definition pattern, but the README doesn't provide an explicit list of all available MCP tools.

Comparable tools

Installation

.package(url: "https://github.com/Tom-Ryder/AgentRunKit.git", from: "2.4.0")
.target(name: "YourApp", dependencies: ["AgentRunKit"])

{
  "mcpServers": {
    "agentrunkit": {
      "command": "swift",
      "args": ["run", "agentrunkit"],
      "env": {}
    }
  }
}

FAQ

What MCP features does AgentRunKit provide?

AgentRunKit includes an MCP client implementation with stdio transport, tool discovery, and JSON-RPC communication capabilities.

Can it run on-device without cloud dependencies?

Yes, it supports on-device inference via MLX on Apple Silicon and Apple Foundation Models.

Compare AgentRunKit with