Class AgenticRagExtensions

Namespace

Mythosia.AI.Rag

Assembly

Mythosia.AI.Rag.dll

Extension methods for enabling Agentic RAG on AIService. Registers the RAG pipeline as a callable tool within the agent's ReAct loop.

public static class AgenticRagExtensions

Inheritance

object

AgenticRagExtensions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Methods

WithAgenticRag<TService>(TService, RagStore, string, string?)

Registers the RAG pipeline as a search tool for use with RunAgentAsync.

In Agentic RAG mode the agent autonomously decides when to search and what query to use. The RagStore's QueryRewriter is intentionally bypassed for this tool — the agent itself is responsible for formulating a clear, self-contained query.

The configured retrieval strategy (vector-only or hybrid) and pipeline options of the RagStore are respected as-is.

public static TService WithAgenticRag<TService>(this TService service, RagStore ragStore, string toolName = "search_documents", string? toolDescription = null) where TService : IAIService, IFunctionRegisterable

Parameters

service TService

The AI service to register the RAG tool on.

ragStore RagStore

A pre-built RAG store containing the indexed documents.

toolName string

The function name exposed to the LLM. Defaults to search_documents. Must be unique among all registered functions on the service.

toolDescription string

Custom description for the tool. When null a sensible default is used. A good description is critical: it controls when the agent decides to invoke this tool.

Returns

TService

The same service instance for fluent chaining.

Type Parameters

TService

A concrete AI service that implements both IAIService and IFunctionRegisterable (e.g. AnthropicService, OpenAIService).

Examples

var ragStore = await RagStore.BuildAsync(cfg => cfg
    .AddDocument("manual.pdf")
    .UseOpenAIEmbedding(apiKey));

var answer = await new AnthropicService(apiKey, http)
    .WithAgenticRag(ragStore)
    .RunAgentAsync("환불 정책을 요약해 줘");