Merge pull request #13 from wisedev-code/feat/add_tool_support

feat: include tool support

Author: wisedev-pstach

src/assets/docs/contracts/stepbuilder.md

@@ -0,0 +1,180 @@
+# ⚙️ StepBuilder Overview
+
+The **`StepBuilder`** class provides a fluent interface for defining the **behavioral flow of an AI Agent**.
+It allows developers to specify **how an agent should process input, use memory, access knowledge, or perform role-based actions** — all through a clear, chainable syntax.
+
+This builder is primarily used when creating or configuring an Agent (e.g., via `AIHub.Agent()`), to define the ordered sequence of reasoning or operational steps the Agent should follow.
+
+---
+
+### 🧩 Namespace
+
+```csharp
+using MaIN.Domain.Entities.Agents.Knowledge;
+using MaIN.Services.Services.Models.Commands;
+```
+
+---
+
+### 🏗️ Class Definition
+
+```csharp
+public class StepBuilder
+```
+
+A utility class for constructing a list of predefined **agent execution steps**.
+Each step defines a mode of operation that influences how the Agent interprets, searches, or responds to input.
+
+---
+
+### 🔹 Core Methods
+
+#### **`Instance`**
+
+```csharp
+public static StepBuilder Instance => new();
+```
+
+Provides a new instance of `StepBuilder` for fluent chaining.
+Used as an entry point for defining a step sequence:
+
+```csharp
+StepBuilder.Instance.Answer().Build();
+```
+
+---
+
+#### **`Answer()`**
+
+Marks a basic answering step — the Agent simply generates a direct response to the input.
+
+---
+
+#### **`AnswerUseMemory()`**
+
+Enables the Agent to **access and use stored memory** when forming a response.
+
+---
+
+#### **`AnswerUseKnowledge(bool alwaysSearchMemory = false)`**
+
+Allows the Agent to **search its knowledge base** to form an answer.
+The optional `alwaysSearchMemory` flag ensures memory is always consulted alongside knowledge retrieval.
+
+---
+
+#### **`AnswerUseKnowledgeWithTags(params string[] tags)`**
+
+Filters knowledge retrieval by **specific tags**, letting the Agent focus on a targeted domain of stored information.
+
+Example:
+
+```csharp
+StepBuilder.Instance.AnswerUseKnowledgeWithTags("finance", "market");
+```
+
+---
+
+#### **`AnswerUseKnowledgeWithType(KnowledgeItemType type)`**
+
+Restricts knowledge usage to a particular **type** (e.g., documents, facts, summaries).
+
+---
+
+#### **`AnswerUseKnowledgeAndMemory()`**
+
+Combines both **knowledge-based reasoning** and **memory recall** for more contextually rich responses.
+
+---
+
+#### **`AnswerUseKnowledgeAndMemoryWithTags(params string[] tags)`**
+
+A hybrid mode using both memory and tagged knowledge retrieval.
+
+---
+
+#### **`Become(string role)`**
+
+Instructs the Agent to **adopt a specific role or persona** before continuing execution.
+
+Example:
+
+```csharp
+.Become("assistant_expert")
+```
+
+---
+
+#### **`FetchData(FetchResponseType fetchResponseType = FetchResponseType.AS_Answer)`**
+
+Defines a step where the Agent **fetches external data**, optionally as a system-level operation.
+
+* `AS_Answer` → Fetches data to produce a normal response
+* `AS_System` → Fetches data for internal processing or state updates
+
+---
+
+#### **`Mcp()`**
+
+Adds an **MCP (Multi-Component Process)** step — typically used for complex, multi-agent or multi-step coordination workflows.
+
+---
+
+#### **`Redirect(string agentId, string output = "AS_Output", string mode = "REPLACE")`**
+
+Redirects execution flow to another Agent.
+Useful for **delegation or agent chaining**.
+
+Example:
+
+```csharp
+.Redirect("knowledge_agent", "AS_Output", "MERGE");
+```
+
+---
+
+#### **`Build()`**
+
+Finalizes and returns the constructed list of steps.
+This list is then passed to the Agent configuration pipeline.
+
+Example:
+
+```csharp
+.WithSteps(StepBuilder.Instance
+    .AnswerUseKnowledgeAndMemory()
+    .Build())
+```
+
+---
+
+### 🧠 Example Usage
+
+```csharp
+var steps = StepBuilder.Instance
+    .AnswerUseKnowledgeAndMemoryWithTags("research", "analysis")
+    .Become("data_researcher")
+    .Build();
+
+await AIHub.Agent()
+    .WithModel("claude-sonnet-4-5-20250929")
+    .WithSteps(steps)
+    .CreateAsync(interactiveResponse: true);
+```
+
+---
+
+### 🔍 Summary
+
+| Category               | Purpose                                       |
+| ---------------------- | --------------------------------------------- |
+| **Answer**             | Simple or knowledge/memory-enhanced responses |
+| **Knowledge & Memory** | Access stored or learned context              |
+| **Role / Behavior**    | Define how the agent behaves or identifies    |
+| **Data Fetching**      | Integrate external or internal data sources   |
+| **Redirection**        | Delegate tasks between agents                 |
+| **MCP**                | MCP server call                               |
+
+---
+
+The `StepBuilder` provides a **modular, declarative way to control agent logic**, making agent configurations expressive, reusable, and easy to reason about.

src/assets/docs/contracts/toolbuilder.md

@@ -0,0 +1,193 @@
+# 🧰 ToolsConfigurationBuilder Overview
+
+The **`ToolsConfigurationBuilder`** class provides a structured and extensible way to define and register **tools** that can be used by AI agents during execution.
+Tools act as **callable functions** — external operations the model can invoke dynamically to retrieve data, perform computations, or trigger actions.
+
+This builder simplifies tool creation by offering multiple overloads for different execution patterns (sync/async, typed/untyped, parameterized/no-parameter tools).
+
+---
+
+### 🧩 Namespace
+
+```csharp
+using System.Text.Json;
+using MaIN.Domain.Entities.Tools;
+```
+
+---
+
+### 🏗️ Class Definition
+
+```csharp
+public class ToolsConfigurationBuilder
+```
+
+A fluent builder for constructing a **`ToolsConfiguration`** object, which contains a list of **`ToolDefinition`** entries and optional configuration settings (such as tool selection behavior).
+
+Each added tool defines a function name, description, expected parameters, and an execution delegate.
+
+---
+
+### 🔹 Core Usage
+
+Used during agent configuration to specify tools the model can access:
+
+```csharp
+.WithTools(new ToolsConfigurationBuilder()
+    .AddTool(
+        name: "get_weather",
+        description: "Retrieve current weather information",
+        parameters: new { type = "object", properties = new { city = new { type = "string" } } },
+        execute: WeatherTools.GetWeather)
+    .WithToolChoice("auto")
+    .Build())
+```
+
+---
+
+### 🔧 Methods Overview
+
+#### **`AddDefaultTool(string type)`**
+
+Registers a basic tool definition with a specific type.
+Used when predefined tool behavior is already encapsulated elsewhere.
+
+```csharp
+.AddDefaultTool("system_logger")
+```
+
+---
+
+#### **`AddTool(string name, string description, object parameters, Func<string, Task<string>> execute)`**
+
+Registers an asynchronous tool that receives raw JSON input and returns a serialized string result.
+This version is suitable when working directly with serialized arguments.
+
+---
+
+#### **`AddTool(string name, string description, object parameters, Func<string, string> execute)`**
+
+Registers a synchronous version of the above method.
+The provided function executes immediately and returns a plain string result.
+
+---
+
+#### **`AddTool<TArgs>(string name, string description, object parameters, Func<TArgs, Task<object>> execute)`**
+
+Registers a **typed asynchronous tool**.
+The input JSON is automatically deserialized into a strongly-typed object (`TArgs`), and the result is serialized back to JSON.
+
+Example:
+
+```csharp
+.AddTool<CreateNoteArgs>(
+    "create_note",
+    "Create a new note with a title and content",
+    new
+    {
+        type = "object",
+        properties = new
+        {
+            title = new { type = "string", description = "Title of the note" },
+            content = new { type = "string", description = "Content of the note" }
+        },
+        required = new[] { "title", "content" }
+    },
+    NoteTools.CreateNoteAsync)
+```
+
+---
+
+#### **`AddTool<TArgs>(string name, string description, object parameters, Func<TArgs, object> execute)`**
+
+Registers a **typed synchronous tool**, similar to the previous method but without asynchronous execution.
+
+The input is deserialized into the specified argument type (`TArgs`), the tool executes synchronously, and the result is serialized to JSON.
+
+---
+
+#### **`AddTool(string name, string description, Func<Task<object>> execute)`**
+
+Registers an **asynchronous parameterless tool**, ideal for actions that don’t require inputs (e.g., fetching current time or status).
+
+Example:
+
+```csharp
+.AddTool("get_current_time", "Get the current date and time", Tools.GetCurrentTimeAsync)
+```
+
+---
+
+#### **`AddTool(string name, string description, Func<object> execute)`**
+
+Registers a **synchronous parameterless tool**.
+This overload is ideal for lightweight, immediate-return tools.
+
+---
+
+#### **`WithToolChoice(string choice)`**
+
+Specifies the **tool selection strategy** for the agent.
+Common values:
+
+* `"auto"` → The model decides when to use tools automatically.
+* `"none"` → Disables tool invocation.
+* `"required"` → Forces tool usage if applicable.
+
+---
+
+#### **`Build()`**
+
+Finalizes the configuration and returns a fully constructed **`ToolsConfiguration`** object.
+
+---
+
+### 🧠 Example Usage
+
+```csharp
+var toolsConfig = new ToolsConfigurationBuilder()
+    .AddTool<GetWeatherArgs>(
+        name: "get_weather",
+        description: "Get current weather conditions for a city",
+        parameters: new
+        {
+            type = "object",
+            properties = new
+            {
+                city = new { type = "string", description = "City name" }
+            },
+            required = new[] { "city" }
+        },
+        execute: WeatherTools.GetWeatherAsync)
+    .AddTool(
+        name: "get_current_time",
+        description: "Return the current local time",
+        execute: Tools.GetCurrentTime)
+    .WithToolChoice("auto")
+    .Build();
+```
+
+This configuration can then be attached to a chat or agent:
+
+```csharp
+await AIHub.Chat()
+    .WithModel("gpt-5-nano")
+    .WithTools(toolsConfig)
+    .CompleteAsync(interactive: true);
+```
+
+---
+
+### 📘 Summary
+
+| Category          | Description                                                                            |
+| ----------------- | -------------------------------------------------------------------------------------- |
+| **Purpose**       | Defines callable tools the model can execute during interaction                        |
+| **Supports**      | Synchronous & asynchronous tools, typed & untyped execution                            |
+| **Serialization** | Automatically handles JSON deserialization for input and serialization for output      |
+| **Integration**   | Used directly with agents or chat sessions via `.WithTools()`                          |
+| **Flexibility**   | Allows parameterized tools, parameterless tools, and dynamic tool selection strategies |
+
+---
+
+The `ToolsConfigurationBuilder` provides a **clean, declarative, and extensible interface** for integrating external logic into AI-driven workflows, enabling agents to **go beyond text generation** and interact with real-world data and actions seamlessly.