You don’t understand how insane this leak is.

Someone leaked the full source code of Claude Code, nearly 2,000 files of Anthropic’s internal codebase.

The code reveals every feature they're building and testing before public release: voice mode, scheduled autonomous agents, multi-agent orchestration, proactive behavior, persistent memory, and - I swear I’m not making this up - a Tamagotchi-style companion pet with gacha rarity tiers.

It’s all wired together under a single vision codenamed Kairos (Greek for “the right moment”), which turns Claude Code from a tool you open and close into a perpetual, always-on AI colleague that, like you, dreams at night to consolidate its memories.

I went through it all (with the help of Claude itself, which makes this a sort of predictive introspection exercise) and translated the code into simple English.

Long story short: Anthropic is building the vertically integrated version of OpenClaw. Here are all the details you need to know.