Description
Add first-class Foundry IQ support to the Agent Framework .NET SDK, equivalent to the Python SDK's AzureAISearchContextProvider agentic mode.
The Python integration supports Azure AI Search Knowledge Bases through a framework context provider, including agentic retrieval, query planning, multi-hop retrieval, extractive or synthesized output, reasoning-effort configuration, and use of either an existing Knowledge Base or one created from an index. In .NET, developers currently need to use lower-level Azure AI Search tooling or build a custom adapter around TextSearchProvider; neither provides the same first-class Foundry IQ integration.
Expected behavior:
- Provide an idiomatic .NET context provider for Foundry IQ/Azure AI Search Knowledge Bases.
- Support agentic retrieval against an existing Knowledge Base.
- Expose relevant retrieval settings such as output mode, reasoning effort, and result limits where supported by the underlying Azure SDK.
- Integrate directly with
AIAgent through AIContextProviders.
- Include documentation and an end-to-end sample corresponding to the Python agentic retrieval sample.
Python references:
Code Sample
var searchProvider = new AzureAISearchContextProvider(new()
{
Endpoint = new Uri(searchEndpoint),
Credential = credential,
Mode = AzureAISearchMode.Agentic,
KnowledgeBaseName = knowledgeBaseName,
OutputMode = KnowledgeBaseOutputMode.ExtractiveData,
RetrievalReasoningEffort = RetrievalReasoningEffort.Minimal,
});
AIAgent agent = chatClient.AsAIAgent(new ChatClientAgentOptions
{
AIContextProviders = [searchProvider],
});
The exact API shape can follow existing .NET naming and options conventions; the important requirement is a supported first-class integration rather than requiring each application to implement its own search adapter.
Language/SDK
.NET
Description
Add first-class Foundry IQ support to the Agent Framework .NET SDK, equivalent to the Python SDK's
AzureAISearchContextProvideragentic mode.The Python integration supports Azure AI Search Knowledge Bases through a framework context provider, including agentic retrieval, query planning, multi-hop retrieval, extractive or synthesized output, reasoning-effort configuration, and use of either an existing Knowledge Base or one created from an index. In .NET, developers currently need to use lower-level Azure AI Search tooling or build a custom adapter around
TextSearchProvider; neither provides the same first-class Foundry IQ integration.Expected behavior:
AIAgentthroughAIContextProviders.Python references:
Code Sample
The exact API shape can follow existing .NET naming and options conventions; the important requirement is a supported first-class integration rather than requiring each application to implement its own search adapter.
Language/SDK
.NET