OpenClaw outlines security roadmap for agent filesystem, network, plugin, command, and static-analysis controls
OpenClaw has published a security roadmap for making a powerful local AI assistant runtime more bounded, observable, and auditable while preserving its ability to act on a user’s machine. The filesystem work centers on `fs-safe`, a shared library for root-bounded filesystem operations. The library is designed to prevent path traversal, absolute-path writes, symlink boundary breaks, […]
OpenClaw