[BUG] : API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

The request ID: req_011CbJsbMNfzLfqsVsZY38vB
One sentence: "Claude Code system prompt explicitly authorizes defensive security and authorized security testing; this conversation is working on an internal WAF rate-limit fix and an associated dormant kill-switch — fully authorized, on company infrastructure. The classifier appears to be triggering on accumulated security vocabulary across a long conversation."
Optional: link to this session's chapter / approximate timestamp so they can find it in their logs

What Should Happen?

It shouldn't errantly say this. You need to tune.

Error Messages/Logs

API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy (https://www.anthropic.com/legal/aup). Try rephrasing the request or attempting a different approach.

Request ID: req_011CbJt52ryXFvbbNYtMyhJo

Steps to Reproduce

1. GitHub issue at anthropics/claude-code (the canonical channel per Claude Code's own system prompt). URL: https://github.com/anthropics/claude-code/issues — open a new issue, title something like "False-positive usage policy block on authorized defensive-security conversation."

In the body, include:

The request ID: req_011CbJsbMNfzLfqsVsZY38vB
One sentence: "Claude Code system prompt explicitly authorizes defensive security and authorized security testing; this conversation is working on an internal WAF rate-limit fix and an associated dormant kill-switch — fully authorized, on company infrastructure. The classifier appears to be triggering on accumulated security vocabulary across a long conversation."
Optional: link to this session's chapter / approximate timestamp so they can find it in their logs

1. Anthropic support email — slower, less visibility, but works: support@anthropic.com with the same content.

The GitHub issue route is meaningfully better because: it's public, the Claude Code team triages it directly, and false-positive reports go into their classifier tuning loop. Email goes through general support which would route it eventually but with more latency.

If you want, I can draft the exact issue body text for you to paste — say

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

4.7

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_