AWS Launches Agent Registry in Preview to Govern AI Agent Sprawl Across Enterprises launches a preview Agent Registry to centrally discover, govern, and reuse enterprise AI agents with native MCP and A2A protocol support. Outcome engineers gain a canonical control plane for agent discovery, versioning, and policy enforcement — a practical step toward Orchestration (Principle 09) and governance (Principle 10).

Anthropic Introduces Agent-Based Code Review for Claude Code rolls out agent-based Code Review that runs parallel AI reviewers to find, verify, and prioritize pull-request issues in roughly 20 minutes. This lets teams embed automated reviewers into CI pipelines to scale review velocity while keeping humans in the loop, a clear Teamwork (Principle 03) and Orchestration (Principle 09) pattern.

Agents That Remember: Introducing Agent Memory launches a managed retrieval-based persistent memory beta so agents can recall context without filling model context windows. Persistent memory reshapes how you design long-running agent workflows, enabling stateful agents, auditability, and cheaper context engineering (Principles 06 and 11).

Introducing Flagship: Feature Flags Built for the Age of AI brings OpenFeature-based, edge-native feature flags to Workers so agents can deploy, test, and roll out code safely without constant human oversight. Feature flags let outcome engineers gate capabilities, run canaries, and quickly roll back agent behaviors — essential operational controls aligned with Gate (Principle 15).

First Take on CadenceLive and Its AI Agent Stacks for EDA unveils a hierarchical agentic EDA stack with a head orchestrator that automates and coordinates system-design workflows. It provides a concrete template for building role-specialized agent pipelines in complex domains, showing Orchestration (Principle 09) and clearer Legible Landscapes (Principle 06) in practice.