[Bug] Anthropic API Safeguard False Positive on Authorized Security Testing

Status Open
Reported on v2.1.220
Maintainer reply None cached
Activity 0 comments · opened Aug 1, 2026

Bug Description

Claude Code’s real-time cyber safeguards interrupted authorized defensive testing of an authentication flow in a backend owned by our organization.

The test was limited to confirming that invalid authentication inputs were rejected. The endpoint behaved correctly, but the safeguard prevented completion and documentation of the negative test suite.

Please review this as a possible false positive during authorized defensive security testing.

Environment

  • Platform: macOS
  • Terminal: Cursor
  • Claude Code version: 2.1.220

View original on GitHub ↗