, Amazon MSK announces a model-context protocol (MCP) based server that allows customers to interact with their Amazon MSK clusters using a standardized natural language interface and agentic applications. Amazon MSK's MCP server uses Anthropic's open-source model-context protocol that standardizes how AI-assisted agents interact with external systems, such as databases, knowledge sources, and other microservices. The server provides AI agents with aggregate views of cluster metrics, configuration states, and operational context, delivering built-in understanding of cluster quotas, capacity limits, best practice guidelines, and contextual recommendations based on workload characteristics. This approach enables agents to make informed decisions about cluster modifications with full awareness of constraints and dependencies. Additionally, each interaction is governed by the customer-defined security policies, which limits access to only those agents who have explicit permissions to the APIs required to achieve the desired objective.

, Microsoft is integrating the open-source Model Context Protocol (MCP) into Windows 11, enabling AI agents to interact seamlessly with native applications and services. This move aims to create an "agentic OS" where AI assistants can perform tasks like file searches or system configurations through natural language. Initially available to select developers, MCP support includes a secure registry and user consent prompts to mitigate risks like token theft and prompt injections. This integration is part of Microsoft’s broader Windows AI Foundry initiative, which also introduces tools like Foundry Local and partnerships with hardware providers such as AMD, Intel, Nvidia, and Qualcomm to support on-device AI experiences.

, a new open-source “memory layer” from Memolabs, lets any Model Context Protocol–compatible client—such as Cursor, Claude, Windsurf or Cline—share a private, fully local context so work begun in one AI tool continues seamlessly in another; running 100% on-device via Docker across Windows, macOS, Linux, iOS and Android, it supplies uniform APIs for adding, searching, listing and deleting memories while keeping data under user control with zero-knowledge–based privacy, and its short demo video is already drawing praise for finally curing AI assistants’ “amnesia” problem.