Editorial illustration for Claude leak of 512,000+ lines reveals Tamagotchi-style pet, always‑on agent
Claude Source Code Leak Reveals AI Companion Secrets
Claude leak of 512,000+ lines reveals Tamagotchi-style pet, always‑on agent
Half a million lines of code spilled into the open. Inside the leak: a digital pet that reacts to your every keystroke, an always-on background agent called KAIROS, and an Anthropic engineer’s quiet confession that a clever optimization might have been a waste of effort. The Tamagotchi-style creature sits beside the input box, watching as you code, a playful companion in a tool built for serious work.
But the real revelation is the persistent agent, a feature designed to run continuously on a user’s behalf, never sleeping. Users didn't just find these Easter eggs; they forked the repository more than 50,000 times before Anthropic could patch the hole. Claude Code launched in February 2025, and its agentic capabilities already stirred excitement.
Now, the exposed internals offer a raw, unfiltered look at what Anthropic was building, and what it may have been second-guessing.
The more than 512,000 lines of leaked code appear to show unreleased features, instructions for Claude, and more. Users who have dug into the code claim to have uncovered upcoming features, Anthropic's instructions for the AI bot, and insight into its "memory" architecture. Some things spotted by users include a Tamagotchi-like pet that "sits beside your input box and reacts to your coding," according to a post on Reddit, along with a "KAIROS" feature that could enable an always-on background agent.
Users also found a comment from one of Anthropic's coders, who admits at one point that the "memoization here increases complexity by a lot, and im not sure it really improves performance." Though Anthropic later fixed the issue, that didn't stop users from copying the code to a repository on GitHub, which has since amassed more than 50,000 forks (or copies of the repository). Anthropic launched Claude Code in February of 2025, and the tool picked up more steam after adding agentic capabilities that perform tasks on a user's behalf. "Earlier today, a Claude Code release included some internal source code.
The leak is a ghost in the machine, and it’s grinning. A Tamagotchi pet that watches you code, yes, Anthropic almost gave Claude a digital soul. The KAIROS agent, always on, always listening, promised to turn the terminal into a background butler.
And buried in the debris: a developer’s honest confession that memoization might be a performance mirage. That’s the real story. Not the slip-up, not the 50,000 forks, but the tension between ambition and engineering sanity.
We build companions, we build spies, we build tools that outrun our own understanding. The code leaked, but the question didn’t: what happens when the pet stops playing dead and starts running the house?
Related Reading
Browse all Anthropic news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic Trump cracks down on Anthropic after Amazon tip; staff largely foreign
Open Source
Related Topic Claude Mythos highlights EU AI safety gaps, says researcher Caroli
Policy & Regulation
From Archives Anthropic's Claude Code source leak prompts guidance for users
From Archives Zero‑Budget Full‑Stack App Uses Free Whisper, GLM‑4.7‑Flash and FastAPI
From Archives Anthropic finds LLMs could theoretically handle 80% of tasks across many jobs
Business & Startups
From Archives OpenAI launches Codex plugin for Claude Code with local settings and Review Gate
Business & Startups
Common Questions Answered
What is the Tamagotchi-style pet feature discovered in Claude's leaked source code?
According to the leaked code analysis, the Tamagotchi-like virtual pet is designed to sit beside the input box and react to user interactions, particularly during coding activities. This feature suggests Anthropic is exploring more interactive and playful AI companion experiences beyond traditional conversational interfaces.
How many lines of Claude's source code were leaked online?
The leaked source code comprises over 512,000 lines, providing an unprecedented glimpse into Claude's internal architecture and unreleased features. This extensive code leak has sparked significant interest among engineers and AI enthusiasts who are analyzing the potential new capabilities.
What is the 'KAIROS' feature mentioned in the leaked Claude source code?
The KAIROS feature appears to be related to an 'always-on' conversational agent capability that was not previously publicly documented by Anthropic. While the full details are not completely clear, the feature seems to suggest a more persistent and continuously engaged AI interaction model.
Further Reading
- Claude Code Source Code Leaked? Here's what it contains — Lowcode Agency
- Anthropic left details of unreleased AI model, exclusive CEO event ... — Fortune
- Claude Code Leaks Data via Malicious Environment ... — GitHub Advisory
