OpenClaw Adds ‘Dreaming’ Memory System, Patches Five Major Security Flaws

OpenClaw is an open-source AI agent platform that allows developers to build agents capable of performing tasks autonomously. Its new “Dreaming” memory system lets agents replay historical notes, organize their experiences and form long-term memories. The feature improves continuity across tasks while drawing greater attention to the secure management of memory data.

OpenClaw’s latest release includes the “Dreaming” memory feature, enabling agents to review past interactions through existing notes and gradually accumulate information for use in future tasks. The platform also patched five critical security vulnerabilities, including risks involving server-side request forgery (SSRF) and environment-variable pollution. Reports did not disclose the exact release date, version number or any associated monetary amount.