Claude Code source code leaks via npm, exposing secret agent features and sparking a supply chain attack

事件摘要

On March 31, 2026, Anthropic accidentally exposed 512,000 lines of Claude Code source code through a source map file in its npm package. The leak revealed hidden features including 'KAIROS' (a persistent background agent), 'Undercover Mode' (for stealth open-source contributions), and 'Dreaming' (autonomous cross-session learning). It also triggered a supply chain attack as threat actors weaponized the leak with trojanized npm packages. The incident became the highest-profile AI source code leak in history.

影响评估

• Risk Creation -1 · Short-term

  Largest AI source code leak in history. Exposed proprietary agent architectures. Triggered supply chain attack with trojanized packages. Established new category of AI security risk: operational code leaks.

  Affected Groups: AI developers, Anthropic users, open-source community, security teams

共识度与来源

• 1
  Claude Code Source Leaked via npm Packaging Error — The Hacker News

  URL: https://thehackernews.com/2026/04/claude-code-tleaked-via-npm-packaging.html

  Claude Code source code leaked via npm packaging error. 512,000 lines of TypeScript files exposed.

  News Report Citation logged Live source
• 2
  Anthropic confirms Claude Code source code inadvertently released — CNBC

  URL: https://www.cnbc.com/2026/03/31/anthropic-leak-claude-code-internal-source.html

  Anthropic confirms Claude Code source code inadvertently released due to human error.

  News Report Citation logged Live source
• 3
  Claude Code Leak: Critical AI Security Threat — Zscaler

  URL: https://www.zscaler.com/blogs/security-research/anthropic-claude-code-leak

  Threat actors weaponizing Claude Code leak with trojanized packages distributing Vidar and GhostSocks.

  News Report Citation logged Live source