Claude Code leak exposes a Tamagotchi-style ‘pet’ and an always-on agent
512,000 lines of Claude Code just leaked. Here's what Anthropic didn't want you to see.
Why it matters
A source map file exposed in Claude Code's 2.1.88 update revealed internal AI instructions, memory architecture, and unreleased features—giving competitors and security researchers rare visibility into Anthropic's coding assistant infrastructure and product roadmap.
The key facts
10 to know512,000+ lines of TypeScript codebase exposed
Leak contained via source map file in Claude Code 2.1.88 update
Unreleased features discovered by reverse engineering
AI bot instructions and memory architecture exposed
Published March 31, 2026
512,000+ lines of TypeScript code exposed via source map file in 2.1.88 update
Leaked data includes unreleased features, system prompts, and memory architecture
Discovery made public on X; reported by Ars Technica and VentureBeat
Published Mar 31, 2026 — timing suggests possible April 1 context; verify authenticity
UNVERIFIED — Claims about Tamagotchi-style 'pet' and always-on agent need corroboration
Go to the source
The Verge AItheverge.com
Publisher excerpt: After Anthropic released Claude Code's 2.1.88 update, users quickly discovered that it contained a package with a source map file containing its TypeScript codebase, with one person on X calling attention to the leak and posting a file containing the code. The leaked data reportedly contains more…