Model Context Protocol ecosystem gains new hosting and testing tools

Developments in the Model Context Protocol (MCP) ecosystem are introducing specialized hosting and testing solutions for AI agents and servers.

MCP Cloud and Manufact Cloud offer distinct hosting models for remote MCP servers. MCP Cloud serves as a centralized control plane for managing inventory, security policies, and cross-server governance. In contrast, Manufact Cloud is designed as a developer-centric platform focused on GitHub-native workflows, allowing product teams to build, preview, and deploy MCP servers and interactive apps through a build-to-production pipeline.

To address the complexities of testing these dynamic interfaces, the bubblemcp-test-kit has been introduced. This lightweight toolkit aims to reduce the boilerplate required for testing MCP servers across various transports, including stdio, HTTP, and SSE. It provides features such as fluent assertions for JSON schemas, mocking capabilities, and agent trace fault injection to ensure AI agents interact with tools correctly.