Microsoft Locks Down Agent Runtimes and the 1.0 Era Begins

Tools

Microsoft Agent Governance Toolkit: Open-Source Runtime Security for AI Agents

Seven packages covering everything from policy enforcement to RL training governance, all intercepting agent actions at sub-0.1ms latency. This is the first toolkit to address all 10 OWASP Agentic AI risks in one place, and the framework-agnostic design means you can drop it into LangChain, CrewAI, or a dozen other stacks without a rewrite.

Codenotary AgentMon Brings Continuous Monitoring to Agentic Networks

AgentMon tracks token usage, communication paths, data access patterns, and secrets handling across multi-agent environments in real time. If you're running agents at any scale and can't answer 'what did my agents touch today,' this fills that gap.

News

Microsoft Agent Framework Hits Version 1.0 with Stable APIs and Long-Term Support

Agent Framework 1.0 unifies Semantic Kernel and AutoGen into a production-ready SDK with stable multi-agent orchestration, A2A and MCP support, and first-party connectors for OpenAI, Anthropic, Gemini, Bedrock, and Ollama. The migration assistants for Semantic Kernel and AutoGen teams remove the main excuse for not switching.

Analysis

Why the Agent Governance Toolkit's MCP Security Scanner Matters Right Now

As MCP becomes the default way agents connect to tools, the toolkit's scanner catches tool poisoning and typosquatting in MCP servers before they hit production. This is a supply-chain attack vector most teams aren't thinking about yet.