AIAgentSharp 1.0.0

AIAgentSharp

A powerful .NET library for building AI agents with LLM integration, tool execution, and state management.

Features

• 🔄 Re/Act Pattern Support: Full implementation of the Re/Act (Reasoning and Acting) pattern for LLM agents
• 🔧 Function Calling: Support for OpenAI-style function calling when available
• ��️ Tool Framework: Rich tool system with automatic schema generation, validation, and introspection
• 💾 State Persistence: Multiple state store implementations (in-memory, file-based)
• 📊 Real-time Monitoring: Comprehensive event system for monitoring agent activity
• 🔄 Loop Detection: Intelligent loop breaker to prevent infinite loops
• 🔄 Deduplication: Smart caching of tool results to improve performance

Quick Start

using AIAgentSharp;

// Create components
var llm = new OpenAiLlmClient(apiKey, "gpt-4");
var store = new MemoryAgentStateStore();
var tools = new List<ITool> { new ConcatTool(), new GetIndicatorTool() };

// Configure agent
var config = new AgentConfiguration
{
    MaxTurns = 40,
    UseFunctionCalling = true,
    EmitPublicStatus = true
};

// Create and run agent
var agent = new StatefulAgent(llm, store, config: config);
var result = await agent.RunAsync("my-agent", "Your goal here", tools);

Console.WriteLine($"Success: {result.Succeeded}");
Console.WriteLine($"Output: {result.FinalOutput}");

Creating Custom Tools

[ToolParams(Description = "Parameters for weather lookup")]
public sealed class WeatherParams
{
    [ToolField(Description = "City name", Required = true)]
    public string City { get; set; } = default!;
}

public sealed class WeatherTool : BaseTool<WeatherParams, object>
{
    public override string Name => "get_weather";
    public override string Description => "Get current weather for a city";

    public override async Task<object> InvokeTypedAsync(WeatherParams parameters, CancellationToken ct = default)
    {
        // Your weather API logic here
        return new { city = parameters.City, temperature = "22°C" };
    }
}

Installation

dotnet add package AIAgentSharp

Documentation

For detailed documentation, examples, and API reference, visit: https://github.com/erwin-beckers/AIAgentSharp

License

MIT License - see LICENSE file for details.