A surprise leak of Anthropic's Claude Code source code recently offered significant insights into the “vibe-coding” scaffolding the company has built around its proprietary Claude model. Beyond the existing features, observers scrutinizing over 512,000 lines of code across more than 2,000 files also discovered references to disabled, hidden, or inactive functionalities, providing a glimpse into the potential roadmap for future enhancements.
Chief among these revelations is “Kairos,” a persistent daemon designed to operate in the background even after the Claude Code terminal window is closed. This system would employ periodic “<tick>” prompts to regularly assess the need for new actions and incorporate a “PROACTIVE” flag for “surfacing something the user hasn’t asked for and needs to see now.”
Kairos leverages a file-based “memory system” intended for persistent operation across user sessions. A prompt, hidden behind a disabled “KAIROS” flag in the code, elucidates the system’s design to “have a complete picture of who the user is, how they’d like to collaborate with you, what behaviors to avoid or repeat, and the context behind the work the user gives you.”
To organize and consolidate this memory system across sessions, the Claude Code source code references an evocatively named “AutoDream” system. When a user goes idle or explicitly instructs Anthropic to enter a sleep state at the end of a session, the AutoDream system would inform Claude Code that “you are performing a dream—a reflective pass over your memory files.”
This prompt, detailing the AI “dream” process, instructs Claude Code to scan the day’s transcripts for “new information worth persisting,” consolidate this new information to avoid “near-duplicates” and “contradictions,” and prune existing memories that are overly verbose or newly outdated. Claude Code would also be tasked with monitoring for “existing memories that drifted,” an issue previously observed when Claude users attempted to graft memory systems onto their custom harnesses. The overarching goal, as per the prompt, is to “synthesize what you’ve learned recently into durable, well-organized memories so that future sessions can orient quickly.”