Claude Code Leak: What the New Features Mean for Work in 2026

Cracked software box with code pages spilling out and icons for memory, agents, and 24/7 operation above it.

The Claude Code source leak exposed roughly 512,000 lines of internal code, giving us a detailed look at features Anthropic is building into its coding agent. The most significant for business users are persistent memory, multi-agent orchestration, and continuous 24/7 background operation. These features signal that Claude is shifting from a chat tool you prompt […]