#multi_agent — Public Fediverse posts on home.social

hasamba @[email protected] · 2026-09-29 · 08:04 UTC

----------------

🛠️ Tool: Microsoft Agent Framework
===================

Microsoft released an open, multi-language framework for building production-grade AI agents and multi-agent workflows. Microsoft Agent Framework (MAF) supports Python, .NET and Go, providing a consistent foundation for building, orchestrating, and operating agent systems.

Purpose and Capabilities

MAF is designed for teams taking agents from prototype to production. The framework provides orchestration beyond a single prompt or stateless chat loop, with graph-based patterns supporting sequential, concurrent, handoff, and group collaboration workflows. It addresses durability, restartability, observability, governance, and human-in-the-loop control requirements that production agent systems typically need.

Key Features
• Python and C#/.NET Support: Full framework support with consistent APIs across both languages
• Go SDK: Available in a separate repository with its own documentation and samples
• Multiple Agent Provider Support: Various LLM providers with continuous additions
• Middleware: Flexible system for request/response processing, exception handling, and custom pipelines
• Orchestration Patterns: Graph-based workflows supporting sequential, concurrent, handoff, and group collaboration, including checkpointing, streaming, human-in-the-loop, and time-travel
• Foundry Hosted Agents: Deploy agents to Foundry-hosted infrastructure with approximately 2 additional lines of code (new)

Technical Implementation

The framework sits within the Microsoft ecosystem, supporting Microsoft Foundry, Azure OpenAI, OpenAI, and the GitHub Copilot SDK. It provides samples and hosting patterns for both local development and cloud deployment. The Python package is available on PyPI as agent-framework, and the .NET package is on NuGet as Microsoft.Agents.AI.

The orchestration engine uses graph-based workflows that support checkpointing for state persistence, streaming for real-time responses, human-in-the-loop for intervention points, and time-travel for debugging and replay. The middleware layer allows custom processing pipelines between agent components, enabling request/response transformation and exception handling.

Use Cases

MAF fits teams building agents and workflows expected to run in production, particularly those needing orchestration beyond simple chat interactions. The graph-based patterns support complex multi-agent scenarios where agents collaborate, hand off tasks, or operate concurrently. Provider flexibility allows teams to switch or add LLM providers without major architectural rewrites.

Considerations

The source is a GitHub README without independent verification. No benchmark data, performance characteristics, or production case studies are documented. Go SDK support is maintained in a separate repository. The ecosystem leans toward Microsoft infrastructure, though external provider support is expanding.

References
• Repository: github.com/microsoft/agent-framework
• Go SDK: github.com/microsoft/agent-framework-go
• PyPI: pypi.org/project/agent-framework/
• NuGet: Microsoft.Agents.AI

🔹 tool #MAF #AI #orchestration #multi_agent

🔗 Source: https://github.com/microsoft/agent-framework

#multi_agent #orchestration #ai #maf