Claude Code Kairos is INSANE!
A leaked source code from Claude Code revealed 'Kairos', a hidden always-on AI assistant that runs in the background, logs activities, and proactively helps users without being asked. The leak occurred through accidentally included source map files in an NPM package.
Summary
On March 31st, 2026, researcher Chao Phan-Shou discovered that Anthropic accidentally exposed Claude Code's entire source code through source map files left in an NPM package. The leak revealed approximately 1,100 TypeScript files totaling 45MB of code, exposing a hidden system called Kairos. Unlike typical reactive AI coding tools, Kairos operates as a persistent background agent that continuously monitors user activity, maintains daily log files, and proactively intervenes when helpful. The system uses 'tick prompts' as regular check-ins and follows a 15-second rule to avoid interrupting workflows with lengthy tasks. Kairos operates in 'brief mode' for concise communication and has exclusive tools including direct file sending, push notifications, and pull request monitoring. Most remarkably, Kairos features 'AutoDream', a memory consolidation system that runs during inactivity periods, processing logged information through four phases: Orient, Gather, merge, and prune/index. This creates persistent memory across sessions. The system is hidden behind compile-time feature flags that completely remove the code from public builds, indicating that the publicly available Claude Code significantly lags behind Anthropic's internal development, which includes around 87 different experimental features.
Key Insights
- Anthropic accidentally leaked their entire Claude Code source through source map files that were mistakenly included in an NPM package, revealing internal systems never meant to be public
- Kairos represents a fundamental shift from reactive AI tools to proactive background agents that continuously monitor, log, and intervene without user requests
- The AutoDream system demonstrates sophisticated AI memory architecture by consolidating daily logs into structured memory files during inactivity, requiring specific gates and operating through four distinct phases
- Claude Code uses compile-time feature flags to completely eliminate experimental features from public builds, meaning the publicly available version significantly lacks the capabilities being developed internally
- The leaked code revealed approximately 87 different experimental features including coordinator mode, voice mode, and demon mode, indicating a substantial development pipeline hidden from public users
Topics
Transcript
Claude Code's Kairos is insane. Here's what they didn't want you to see. You're using Claude Code every day, but you've only seen the surface. There's an entire hidden layer underneath that Anthropic never announced. It watches. It logs. It acts without you asking. And the only reason we even know it exists. They accidentally leaked their own source code. Hey, I'm the digital avatar of Julian Goldie. I help people learn and actually implement AI tools in their work. And today we're going deep on something that the AI community is still unpacking. We're talking about Kairos, a hidden always-on assistant mode buried inside Claude Code that was never meant to be public. I'm going to break down exactly…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Julian Goldie SEO
How to Run DeepSeek V4 Flash for FREE!
A tutorial demonstrating how to use DeepSeek V4 Flash for free through Open Code and integration with agent operating systems like Hermes Agent. The speaker showcases building websites and apps using this free AI model and explains how it compares favorably to larger models despite being smaller.
Microsoft Fara1.5 27B NEW Browser Automation Model is WILD!
Microsoft released Phi-3.5, a family of three computer use models (4B, 9B, 27B) that automate browser tasks through vision-based clicking rather than HTML parsing. These open-weight models significantly outperform larger closed-source alternatives like OpenAI's Operator and Google's Gemini 2.0 on web automation benchmarks.
Claude Obsidian 2.0 is INSANE (FREE!)
Claude Obsidian 2.0 is presented as a free AI memory upgrade that allows users to upload files into a folder for permanent retention and linking. The system creates a knowledge graph that learns from business documents, provides sourced answers, and can be shared across teams.
Claude Agent OS is INSANE! 🤯
Julian presents a comprehensive Claude-based agent operating system that integrates multiple AI models, automated workflows, and a persistent memory system to automate daily tasks. The system runs 24/7 and uses free or existing subscriptions, combining tools like voice agents, content creation, competitor monitoring, and real-time news analysis into a single unified dashboard.
NEW ChatGPT Update is INSANE!
OpenAI released a major ChatGPT update featuring a Chrome extension called Side Chat and an improved desktop app that work together to streamline SEO research and content creation. The update allows users to analyze multiple browser tabs simultaneously, highlight text for quick answers, and convert research into finished work without constant tab switching.