On November 6th, the Lima project team released version 2.0, expanding its capabilities beyond container management to include AI workflows. Lima, a command line tool for launching local Linux virtual machines, originally aimed to support containerd for Mac users. Since its inception in May 2021, Lima has evolved significantly, now accommodating non-container workloads and non-macOS hosts after joining the CNCF as a Sandbox project in September 2022 and reaching Incubating status in October 2025. Key updates in this release include a plugin infrastructure allowing third-party developers to add features without altering the core Lima code. New plugins include VM driver plugins for additional hypervisors, CLI plugins for enhanced command capabilities, and URL scheme plugins. GPU acceleration has been integrated for krunkit, enabling better performance for AI applications running on macOS. The introduction of Model Context Protocol (MCP) tools allows for secure file operations inside a VM sandbox, providing alternatives to existing command-line tools. Lima's new focus on AI is particularly relevant for running AI coding agents in isolated environments. This setup prevents agents from accessing host files directly, minimizing the risk of executing harmful commands. Users can run AI agents either inside Lima, utilizing its GPU features, or as host processes outside Lima, leveraging MCP tools to monitor file access and command execution. For those looking to run AI applications within Lima, the installation process is straightforward, and specific commands guide users in configuring their virtual environment securely.