Rebound is a free MCP server tool developed by Rebound How, built primarily with Rust. Its fault component functions as an AI MCP server, designed to simulate network disruptions directly from a user's machine. This platform aims to help engineers and organizations test and improve the resilience of their software systems. Rebound supports communication with several AI clients, making it a versatile tool for operations and software delivery teams. It is distributed via a simple shell script.

What does Rebound connect to?

Rebound's fault component connects to various AI clients and models, acting as an MCP server. It simulates network disruptions directly from the user's local machine. Confirmed compatible AI clients include Claude, Replit Agent, OpenAI, Microsoft, and Google DeepMind. The server supports two standard transport mechanisms for communication: stdio (standard input/output) and Streamable HTTP, which replaced an earlier HTTP+SSE transport. This allows for flexible integration with different client setups.

Core features and tools

The fault MCP server is designed to simulate network disruptions. While the specific callable tools for disruption simulation are not explicitly named in the documentation, the Model Context Protocol generally allows clients to discover available tools via a tools/list endpoint and invoke them using a tools/call endpoint. This design enables clients to interact with Rebound's capabilities, even without a predefined list of its internal disruption simulation functions.

Who is it for?

Rebound is intended for engineers and organizations focused on software delivery, operations, and building resilience strategies. It provides a way to test and improve the stability of software systems by simulating adverse network conditions. This program helps teams identify weaknesses and strengthen their applications against real-world failures. Skip this if your work does not involve testing software system resilience or managing operational stability.

How to get started

Access is granted through a standard OAuth authorization flow, as specified by the MCP standard, although many local deployments using stdio transport often default to no caller authentication. For Streamable HTTP connections, servers are advised to validate the Origin header to prevent DNS rebinding attacks and ensure proper authentication is in place, despite OAuth 2.1 setup frequently being omitted in basic deployments.

Use cases

• Simulating network disruptions to test and improve the resilience of software systems is a concrete workflow supported by the Rebound toolbox, including the fault MCP server, for engineers and organizations focused on software delivery, operations, and building resilience strategies, helping them identify vulnerabilities and strengthen their applications against real-world failures.

Rebound, a free MCP server from Rebound How, provides a specialized tool for engineers and organizations to assess and strengthen software system resilience. Its fault component, built with Rust, connects with major AI clients to simulate network disruptions. While specific disruption tools are not named, its support for stdio and Streamable HTTP offers connection flexibility. The program addresses a critical need for testing system stability under adverse conditions.