microsoft-agent-framework

Multi-agent workflows are orchestrated using patterns like GroupChat, Sequential, Concurrent, and Handoff. The framework supports building scalable AI assistants and orchestrating tools in sequence or in parallel across Python and .NET.

Declarative YAML agents define AI agent configurations for repeatable deployments. You need this approach when you want to standardize agent behaviors and tool integration across different environments without writing imperative code.

Yes, you can manage memory and state for chat agents. The framework provides ChatAgent capabilities with built-in memory and state management to maintain context across multi-agent workflows and tool integration scenarios.

You integrate external tools with AI agents using the @ai_function decorator. This allows your ChatAgent to call Python or .NET functions directly, enabling tool orchestration within single-agent and multi-agent workflows.

Yes, the framework provides observability features to monitor agent health and behaviors. This allows you to track agent performance and operational status during tool orchestration and multi-agent workflows.

The best way to build a single-agent chat application is using the ChatAgent component with @ai_function decorators for tool integration. This setup supports memory management and observability for scalable AI assistants.