## Microsoft Launches "Wassette" for Secure AI Agents Microsoft has unveiled a new open-source project called **Wassette**, designed to enhance the security of AI agents. This project, lice…
## Microsoft Launches "Wassette" for Secure AI Agents Microsoft has unveiled a new open-source project called **Wassette**, designed to enhance the security of AI agents. This project, licensed under MIT, leverages Rust and WebAssembly technologies. Wassette aims to provide a secure environment for AI agents by utilizing the power of WebAssembly, similar to the sandboxing found in web browsers.
### Key Features and Functionality: * **Built on Wasmtime:** Wassette relies on the Wasmtime runtime for execution. * **OCI Registry Integration:** Allows AI agents to fetch and execute WebAssembly components from Open Container Initiative (OCI) registries autonomously. * **Enhanced Security:** Designed with a "deny-by-default" approach for added safety controls.
* **Rust-Based:** Written in Rust, Wassette avoids run-time dependencies. * **Model Context Protocol (MCP) Compatibility:** Works with any AI agent supporting the MCP. * **Cross-Platform Support:** Compatible with Linux, macOS, and Windows operating systems. ### Purpose and Goal The main goal of Wassette is to provide a secure and controlled execution environment for AI agents, mitigating potential security risks.
By utilizing WebAssembly, Microsoft aims to offer a robust and isolated execution environment.