An accidental source map leak exposed the full Claude Code CLI source, revealing anti-distillation fake tool injection, "frustration regexes" tracking user sentiment, and an undercover mode masking Claude Code's identity in API requests. The leak is Anthropic's second accidental exposure in a week, coming just after the model spec leak and amid legal action against OpenCode for API misuse. The findings illuminate Anthropic's defensive strategies against model distillation and third-party Claude subscription abuse.
Products
The Claude Code Source Leak: fake tools, frustration regexes, undercover mode
Anthropic's source map leak exposed Claude Code's anti-distillation defenses including fake tool injection, sentiment tracking, and identity-masking, marking the second major accidental code exposure in a week.
Tuesday, March 31, 2026 12:00 PM UTC2 MIN READSOURCE: Hacker NewsBY sys://pipeline
Tags
products
/// RELATED
ResearchApr 28
An Intelligent Fault Diagnosis Method for General Aviation Aircraft Based on Multi-Fidelity Digital Twin and FMEA Knowledge Enhancement
Researchers combine multi-fidelity digital twins with FMEA knowledge to train ML models for automated fault diagnosis in general aviation, reducing unscheduled maintenance events.
ModelsApr 28
BiTA: Bidirectional Gated Recurrent Unit-Transformer Aggregator in a Temporal Graph Network Framework for Alert Prediction in Computer Networks
BiTA, a hybrid GRU-Transformer architecture, improves network attack detection by capturing temporal dependencies in alert patterns—outperforming existing temporal graph models.