Microsoft has announced that the Microsoft Agent Framework has reached Release Candidate status for both .NET and Python. This milestone indicates that the API surface is stable and feature-complete for what is planned in version 1.0, setting the stage for an upcoming general availability release. For developers building AI-powered assistants or complex agentic systems, this release is a significant step toward a unified, production-ready toolset.
The Microsoft Agent Framework is an open-source development framework designed to build, orchestrate, and deploy AI agents with a consistent programming model across .NET and Python. It succeeds earlier efforts such as Semantic Kernel and AutoGen, consolidating agent creation, orchestration primitives, and multi-provider support under a single SDK. The framework supports common patterns for creating autonomous agents as well as workflows that combine multiple agents, and it integrates with a variety of AI model providers.
Before this release candidate, developers experimenting with Microsoft’s agent technologies had to piece together capabilities using Semantic Kernel or experimental multi-agent orchestrators. Those tools provided early building blocks for agent creation and function invocation, but lacked a stable, unified API suitable for enterprise-grade systems. With this RC release, the framework’s APIs and workflows are locked down, allowing teams to start production evaluation and implementation with greater confidence.
The framework emphasizes simplicity and flexibility. Developers can create a basic AI agent in just a handful of lines in either Python or .NET, using client libraries to connect to various model providers.
using System.ClientModel.Primitives;
using Azure.Identity;
using Microsoft.Agents.AI;
using OpenAI;
using OpenAI.Responses;
var agent = new OpenAIClient(
new BearerTokenPolicy(new AzureCliCredential(), "https://ai.azure.com/.default"),
new OpenAIClientOptions() { Endpoint = new Uri("https://.openai.azure.com/openai/v1") })
.GetResponsesClient("gpt-4.1")
.AsAIAgent(name: "HaikuBot", instructions: "You are an upbeat assistant that writes beautifully.");
Console.WriteLine(await agent.RunAsync("Write a haiku about Microsoft Agent Framework."));
Function tools enable agents to call domain code in a type-safe way, and graph-based workflows support sequential and concurrent agent orchestration, including human-in-the-loop scenarios and streaming responses.
Interoperability features add support for standards such as MCP (Model Context Protocol), as well as agent-to-agent communication patterns.
Multi-agent orchestration is a core capability of the framework. Instead of individual agents acting in isolation, developers can define workflows that coordinate multiple agents participating in tasks with handoff logic and group chat patterns. These orchestration primitives come with built-in support for streaming updates and checkpointing, which are critical for real-time and stateful applications.
There are some caveats at this stage. As a release candidate the packages are still flagged as pre-release on NuGet and PyPI, and the framework is evolving rapidly based on early feedback. Full GA documentation and migration guides are being prepared, including guidance for teams migrating from Semantic Kernel or from AutoGen. While the RC status signals stability, developers should plan to stay current with updates until GA, and be prepared for minor breaking changes if needed.
For practitioners ready to explore the Agent Framework now, Microsoft has published examples and getting-started guides on the official documentation site, along with source code and migration aids on GitHub.
