Microsoft releases version 1.0 of its official Model Context Protocol C# SDK, adding enhanced authorization discovery, incremental scope consent, tool calling in sampling, and other key features aligned with the 2025-11-25 MCP specification.
Microsoft has announced the release of version 1.0 of its official Model Context Protocol (MCP) C# SDK, bringing full support for the 2025-11-25 version of the MCP Specification. The new version introduces several notable capabilities aimed at developers building MCP servers and clients in .NET.

One of the most significant additions is enhanced authorization server discovery. Under the updated specification, servers can now expose Protected Resource Metadata through three different methods, offering more flexibility compared to the single method previously required. The SDK handles the full discovery process on the client side automatically.
The release also adds icon support for tools, resources, and prompts. Developers can now assign icons to these components using a simple attribute parameter or configure more advanced options such as multiple icons, MIME types, size hints, and theme preferences. Server and client metadata have similarly been extended to support icons and website URLs.
Incremental scope consent is another key feature in this release. It applies the principle of least privilege to MCP authorization by allowing clients to request only the minimum access needed for each operation. Previously, clients often had to request all possible permissions upfront. With the new mechanism, clients start with minimal scopes and request additional ones as required; the SDK handles this automatically on the client side.
URL mode elicitation has also been introduced, enabling servers to redirect sensitive interactions, such as API key collection or third-party authorization, away from the MCP client entirely and to a secure server-hosted URL. This approach prevents sensitive data from being transmitted through the client.
Tool calling support in sampling represents one of the most technically significant additions. Servers can now include tools in sampling requests, allowing a language model to invoke those tools during a response. The SDK's integration with the Microsoft.Extensions.AI package simplifies the implementation on both client and server sides.
The release further includes OAuth Client ID Metadata Documents (CIMDs) as a preferred alternative to Dynamic Client Registration, and improved support for long-running requests over HTTP through a polling mechanism that allows servers to close connections and clients to reconnect using event IDs.
Other changes in this release include experimental Tasks support for durable state tracking, DistributedCacheEventStreamStore for SSE stream storage, and extended Implementation metadata properties.
For interested developers, the original announcement blog post provides detailed implementation steps and code examples for each feature covered in this release, including authorization configuration, icon setup, incremental scope consent, elicitation handling, tool calling in sampling, and long-running request management. The full release notes and documentation are available on the official .NET Developer Blog, with the C# SDK repository hosted on GitHub.
This release represents a significant step forward for .NET developers working with MCP, providing comprehensive support for the latest protocol features while maintaining the simplicity and integration that .NET developers expect. The enhanced security features, particularly incremental scope consent and URL mode elicitation, address key concerns around authorization and data protection in AI agent workflows.
The timing of this release aligns with growing interest in MCP as a standard for AI agent communication, making it easier for .NET developers to participate in the broader ecosystem of AI-powered applications and services.

Comments
Please log in or register to join the discussion