Cullis Ships Federated Agent Trust Broker and Agent Framework 1.0 Goes Deep on Multi-Agent Workflows

Cullis Ships Federated Trust Broker for Cross-Organization AI Agents

Cullis is a new open-source project building the trust infrastructure layer that lets AI agents from different organizations authenticate and communicate securely using x509 mutual TLS, JWT RS256, and an append-only audit log. This fills a real gap: most agent frameworks handle intra-org orchestration fine but have no answer for agents that need to transact across org boundaries with verifiable identity and explicit policy authorization.

Microsoft Agent Framework 1.0: What the Stable Release Actually Includes

Agent Framework 1.0 locks in stable APIs for multi-agent orchestration patterns including sequential, concurrent, handoff, group chat, and Magentic-One, plus graph-based workflows with checkpointing and human-in-the-loop approvals. If you are still on AutoGen or Semantic Kernel, the migration guides are now live and this is the signal to move.

Microsoft Agent Governance Toolkit: The Full Architecture Explained

The seven-package toolkit maps directly to all 10 OWASP Agentic AI risks and borrows proven patterns from OS kernels, service meshes, and SRE to enforce policy at sub-millisecond latency before any agent action executes. The stateless policy engine design means it scales horizontally without coordination overhead, which is the right call for production agent systems.