Cursor、チームマーケットプレイス機能を更新 Team Marketplace Updates

Cursor, the AI-powered code editor, has announced updates to its Team Marketplace, a feature designed to help organizations share and govern the extensions and Model Context Protocol (MCP) servers used across their developer base. The update appears aimed at strengthening enterprise governance as Cursor pushes deeper into team and organizational deployments.

The Team Marketplace is a centralized mechanism that lets organizations curate the extensions and AI integration tools available to their developers within Cursor. Administrators can present an internally approved catalog of tools, while developers can browse and install from that vetted set with confidence that the entries have been cleared for organizational use. The latest update reportedly improves the user experience around sharing, discovery, and permissions management, with particular attention paid to how MCP servers are handled within the marketplace flow.

MCP is an open specification originally proposed by Anthropic that defines a common interface through which AI assistants can securely access external data sources and tools. Cursor was among the earlier adopters of the protocol and has built out an ecosystem in which the editor can connect to services such as GitHub, databases, and internal company APIs. That openness has been a significant driver of Cursor's appeal to developers, but it has also raised concerns: when individual developers are free to install arbitrary MCP servers, those connections can become unmonitored channels through which sensitive code or credentials may leak. Establishing organization-level visibility and an approval workflow has therefore become an increasingly pressing requirement for enterprise customers.

By consolidating extension and MCP server distribution into a managed marketplace, Cursor is effectively providing administrators with a control plane for AI tooling. Rather than relying on developers to independently evaluate the trustworthiness of each MCP server, security and platform teams can pre-approve entries, surface recommended tools, and presumably restrict or audit what flows through the editor. The update appears to refine these workflows rather than introduce a wholly new architecture, suggesting an iterative tightening of governance primitives that Cursor has been building over recent releases.

Similar moves are visible across the broader AI coding tools market. GitHub Copilot has expanded its enterprise controls around extensions and policy management, and JetBrains AI Assistant has likewise emphasized administrative configuration for organizations. As the center of gravity in AI-assisted development shifts from individual hobbyist usage toward team and enterprise adoption, the depth of governance features is becoming a meaningful axis of competition alongside raw model quality and editor ergonomics. Procurement teams evaluating these tools increasingly ask not only how well the assistant writes code, but also how administrators can constrain its behavior, audit its activity, and ensure that integrations comply with internal security policies.

For Cursor specifically, the Team Marketplace updates fit a pattern of enterprise-oriented features rolled out over the past year, including expanded admin dashboards, usage analytics, and privacy controls. The company has been positioning itself not just as a developer-facing product but as a platform that CIOs and security leaders can sanction at scale. MCP governance is a natural extension of that positioning, given that the protocol's flexibility is simultaneously its greatest strength and its most acute risk surface.

It remains to be seen how granular the new controls are in practice — for example, whether administrators can scope MCP server availability by team, enforce version pinning, or integrate marketplace approvals with existing identity and access management systems. Cursor has not, in the changelog entry itself, detailed every dimension of the update, and some of the practical implications may only become clear as enterprise customers adopt the new flows. Still, the direction of travel is consistent with broader industry expectations: as AI coding assistants embed themselves more deeply into software development workflows, the marketplaces and protocols that connect them to enterprise data will need to be governed with the same rigor as any other piece of production infrastructure.