[Bug] Anthropic API Error: Overly broad content safety filter blocks legitimate security testing workflows

Bug Description
That's a false positive from Claude's real-time cyber safeguards — the classifier on Sonnet 5 (which Claude Code runs on) pattern-matched the offensive-security vocabulary in that prompt (SQLi, attack tab, exploit, pentest) even though what you're doing is entirely legitimate: building a defensive findings-review dashboard for a scanner you operate, against your own Juice Shop test instance. The safeguard is intentionally broad and sometimes catches authorized work — the message says as much. Nothing's broken; it flagged that one message's wording, not the task.

Environment Info

• Platform: win32
• Terminal: windows-terminal
• Version: 2.1.225
• Feedback ID: 396b243d-471f-4af1-97ce-e1d1f96207cb

Errors

[]