MCP Weekly: Cloud Standardization, Security Platforms, and $200M Agent Investment
December 5, 2025

MCP Weekly: Cloud Standardization, Security Platforms, and $200M Agent Investment

Cloud standardization and security shifts powering next-generation agents

Table of Contents

Welcome to the latest installment of the MCP Weekly digest, covering the major developments related to the Model Context Protocol (MCP) from November 27th through December 4th! Agents hit the cloud-architecture layer this week, pushing security to the top of the stack.

TL;DR

AI infrastructure took a major step forward this week as MCP became the backbone for how large enterprises build and secure agent-driven systems.The headline news is AWS solidifying its commitment by launching the AWS MCP Server, IAM-secured endpoint for over 15,000 APIs and the API Gateway MCP Proxy, enabling all existing REST APIs to be instantly agent-accessible.

Simultaneously, the new reality of autonomous agents executing sensitive commands drove a surge in the dedicated AI Action Security market, with Netskope and Salt Security launching critical visibility and protection capabilities, while Helmet Security emerged from stealth with $9M in funding. Finally, Snowflake and Anthropic announced a $200 million strategic partnership to deploy governed, multi-step AI agents (Claude 4.5) across regulated industries, demonstrating the massive capital flowing into secure, agent-powered data operations.

Major Updates of the Week

AWS Standardizes the Cloud with Unified MCP Architecture

AWS delivered three critical services this week, standardizing the deployment, tool access, and governance of agentic workloads on the world’s leading cloud platform.

Product Key Action / Product Significance
AWS MCP Server (Preview) Unifies the AWS API MCP and Knowledge servers into a single, managed remote service, providing IAM-secured access to 15,000+ AWS APIs. Enables AI agents to perform complex, multi-step tasks (via Agent SOPs) like provisioning EC2 or troubleshooting Lambda with full AWS IAM authentication and CloudTrail audit logging.
IAM Policy Autopilot (Open Source) An open-source MCP server that performs deterministic code analysis (Python, Go, TS) to generate reliable, identity-based IAM policies. Integrates with AI coding assistants (like Kiro) to ensure they generate valid, sufficient permissions based on application code, preventing common security misconfigurations.
API Gateway MCP Proxy Adds native MCP proxy support to transform existing REST APIs into MCP-compatible endpoints. Allows organizations to instantly expose their entire existing API catalog to AI agents via protocol translation, eliminating the need for application changes and enabling semantic search for tool discovery.

In a key supporting update, the AWS Knowledge MCP Server also added topic-based search on November 26, 2025, which improves agent accuracy by allowing queries to target specific documentation domains (e.g., Troubleshooting, AWS CDK) to reduce response noise.

The Agent Security Sector Takes Off

The rapid adoption of MCP has confirmed a new threat vector, driving the immediate need for a dedicated security layer. This week saw major platform-level announcements and significant funding for the new Agent Security market.

  • Netskope Model Context Protocol (MCP) Security: Netskope added breakthrough capabilities to its Netskope One platform to secure MCP communications. This provides full visibility over MCP servers and clients, extends its Cloud Confidence Index (CCI) for risk-scoring MCP servers, and enforces granular, context-based policies and real-time Data Leak Prevention (DLP) to protect sensitive data flowing through agent workflows.
  • Salt Security and AWS WAF Integration: Salt Security, the API security leader, extended its behavioral threat protection to the MCP layer. Using its MCP Finder technology to discover unknown servers, Salt now integrates with AWS WAF to automatically detect and block external misuse and abuse of MCP servers by malicious AI agents and attackers, applying intent-based behavioral analysis to stop threats traditional tools miss.
  • Helmet Security Funding: Helmet Security emerged from stealth with $9 million in funding, positioning itself as the first end-to-end security platform for the agent lifecycle. The company addresses the 17,000+ unmonitored MCP servers in the enterprise by providing instant Discovery, Monitoring, and Management through structured access policies, giving CISOs control without hindering developer speed.

Massive Enterprise Tooling and Data Scale

Major software vendors and data platforms reinforced MCP as the protocol for large-scale, governed AI operations.

Vendor Product Key Action / Product Significance
Snowflake and Anthropic $200M Strategic Partnership A multi-year, $200 million commitment to deploy Claude 4.5 models (Opus, Sonnet) directly within the Snowflake Data Cloud. Enables enterprises to run complex, multi-step analysis via governed Snowflake Cortex Agents, ensuring agents operate strictly within the Snowflake Horizon Catalog’s governance and security perimeter for regulated industries.
Atlassian (Open Source) Rovo MCP Server / ChatGPT Connector Launched a connector allowing ChatGPT to securely access Jira and Confluence via the Rovo MCP Server. Provides external AI assistants with a secure “live source of truth” and, critically, enables writeback support (e.g., creating Jira issues) while enforcing OAuth, audit logs, and an administrator-controlled domain allowlist.
Microsoft Foundry MCP Server (Preview) A secure, fully cloud-hosted MCP service with a public endpoint. Simplifies development by providing a curated toolset for managing the AI lifecycle (Agents, Evaluations, Models) and mandates enterprise security via Entra ID (OAuth 2.0) and adherence to the signed-in user's Azure RBAC permissions.

My Thoughts

Last week we celebrated MCP’s one-year anniversary, and it made me realize how quickly things are changing. MCP started as a simple way for models to use tools, but it’s turning into the core layer that lets agents take real action safely. Over the next year, I think MCP will shape how teams design, approve, and trust AI-driven work.

Looking ahead, the biggest changes won’t just come from new tools but they’ll come from how people actually use MCP day-to-day. We’ll see clearer workflows, shared playbooks, and easier ways to track what agents are doing. The future of AI feels less chaotic now, because MCP is becoming the structure that holds it all together.

Om Shree

Technical Evangelist

About Om Shree

Om Shree is a researcher, technical writer, and AI evangelist who focuses on making complex AI and agent workflows easier to understand. Om's passion is  breaking down emerging technologies into clear, practical insights. He's excited to provide useful in-depth research  that supports product planning and helps developers navigate new tools and systems with ease.

Further Reading

Turn Your OpenAPI Specs Into MCP Tools—Instantly
Introducing a powerful new feature in Gentoro that lets you automatically generate MCP Tools from any OpenAPI spec—no integration code required.
April 22, 2025
6 min read

Customized Plans for Real Enterprise Needs

Gentoro makes it easier to operationalize AI across your enterprise. Get in touch to explore deployment options, scale requirements, and the right pricing model for your team.