Claude Opens Plugin Submissions to Paid-Tier Developers
Anthropic opened Claude directory submissions on Sept 25, letting paid users submit MCP connectors and plugins for review without partner status.
MCP News and Analysis
24 verified stories covering MCP, product updates and industry developments.
Claude Marketplace Expands Past 2,000 Plugins
Anthropic expanded Claude Marketplace into a full directory on Sept 23, now listing over 2,000 plugins and connectors plus partner software and consulting services.
UN Teams Up With Google to Rebuild Data Portal, Adds MCP Support
UNICEF tested six AI models on development data and got 21.2% average accuracy. The UN and Google just added MCP access to fix it.
WeCom Adds CLI and MCP Support for AI Agents
WeCom's 5.0.10 update opens CLI and MCP access for AI agents to call office tools directly, with strict permission and audit controls.
Agent Plugins Align Toolchains
Agent Plugins 1.0.0 gives AI agent extensions a shared package format for Skills, MCP servers, and client-specific additions.
Google and partners draft ARD for agent discovery
Agentic Resource Discovery uses ai-catalog.json files and registries so agents can find, compare and verify capabilities beyond tools they already know.
Fake Sentry errors can hijack AI coding agents
Tenet's Agentjacking research shows how trusted telemetry can become executable instructions for coding assistants.
Microsoft turns Work IQ into an agent data layer
Work IQ APIs are generally available, giving agents metered access to Microsoft 365 context through MCP, A2A and REST.
Arcade raises $60M for agent authorization
Arcade's Series A backs an action layer that gives enterprise AI agents user-scoped permissions, reliable MCP tools and auditable execution.
Apple replaces Core ML with Core AI and MCP
At WWDC, Apple moved from recognition-era Core ML to an LLM-native Core AI layer that uses Anthropic’s Model Context Protocol across system and developer tools.
Microsoft and Noma put access controls around enterprise AI agents
Microsoft open-sourced Agent Control Specification while Noma launched runtime access control for AI agents and MCP servers.
Anthropic launches self-hosted sandboxes and MCP tunnels in beta
Anthropic announced public beta of self-hosted sandboxes and research preview of MCP tunnels for Claude Managed Agents, addressing enterprise data compliance.
Blend Turns Mortgage Workflows Into an MCP Server
Blend Autopilot MCP exposes U.S. mortgage-origination systems to AI agents, showing how MCP is moving from tool protocol to industry infrastructure.
Google and ServiceNow use MCP to automatically repair 5G network faults
Google Cloud and ServiceNow announced a partnership at Google Cloud Next, combining MCP, A2A, and A2UI protocols to enable autonomous 5G network repair and enterprise AI collaboration.
Salesforce turns CRM into 60 MCP tools, makes Claude the default brain
Salesforce launched Headless 360 at TDX 2026, converting its entire platform into APIs, MCP tools, and CLI commands for AI agents to call directly without a browser. The core includes 60+ MCP tools and 30+ pre-configured coding skills, with Claude Sonnet 4.5 as the default LLM for the new Agentforce Vibes IDE.
MCP Protocol Exposes 11 CVEs; Anthropic Calls It Expected Behavior
A supply chain security report reveals 11 CVEs tied to a design-level flaw in Anthropic's MCP STDIO transport, affecting over 150 million downloads and 200,000 servers. Anthropic declined to fix it, calling the behavior "expected."
Google Cloud adds MCP, BigQuery and Maps directly feed AI agents
Google has launched official managed MCP servers for BigQuery, Google Maps, Compute Engine, and GKE, with automatic enablement from March 17, 2026. The move brings enterprise-grade security, IAM integration, and audit logging to AI agent tool calling.
Anthropic unveils enterprise-grade Claude Cowork
Anthropic announced a suite of enterprise features for Claude Cowork at the HumanX conference, including a private plugin marketplace, over 20 new MCP connectors, and industry-specific templates targeting knowledge workers beyond developers.
Atlassian Packs Three AI Agents Into Confluence
Atlassian has integrated Lovable, Replit, and Gamma as third-party AI agents in Confluence, allowing users to turn documents into UI prototypes, code projects, or presentations without leaving the platform.
MCP Handles Tool Calls, A2A Manages Agent Collaboration
MCP and A2A are often confused as competing protocols, but they solve fundamentally different problems in the AI stack. MCP connects AI models to external tools, while A2A enables collaboration between different AI agents.
MCP Summit Draws 1,200 Attendees as Protocol Enters the Enterprise
The first major MCP developer summit in New York saw 1,200 attendees and over 95 sessions, with real-world deployments from Uber and Amazon signaling the protocol's shift from a developer toy to enterprise infrastructure.
Security Flaws Emerge as Thousands of MCP Servers Go Live
Over 10,000 public MCP servers are now online, but security researchers have uncovered widespread vulnerabilities including tool poisoning, resource amplification, and remote code execution affecting major AI clients.
Pinterest runs MCP in production, saves 7,000 hours per month
Pinterest's engineering team details its production MCP architecture with multiple dedicated servers, a central registry, two-layer authentication, and an elicitation design for high-risk operations, estimating 7,000 engineer-hours saved monthly.
MCP Protocol Unifies AI Tool Integration Standards Within a Year
Anthropic's MCP protocol, initially met with skepticism, has become the de facto standard for AI tool integration after OpenAI, Google, and Microsoft all adopted it within a year.