False positive: Usage Policy block on technical markdown integration task

What happened

In the middle of a long-running Claude Code session on a personal
project, a routine user message was blocked with:

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_011CbJsgRw9q1h9LjMxgjbnY

Context

The session was integrating section drafts into a markdown
investigation report — a technical document about per-file-type
extraction mechanism selection for email attachments. Nothing in
the conversation has plausible policy relevance (no security/abuse/
sensitive-content angle; ordinary software-engineering work on a
single-user productivity tool).

The blocked turn was a normal continuation of that work — answers
to a prior AskUserQuestion plus a request to proceed. Retrying
was not attempted before filing this; reporting now while the
Request ID is fresh.

Expected behavior

Request handled normally — no Usage Policy intercept on technical
markdown editing tasks.

Actual behavior

Hard block returned to the client; user had to interrupt the
in-progress workflow to investigate.

Repro

Not deterministically reproducible from the user side — the
classifier signal isn't visible. Including the Request ID so the
upstream decision can be inspected.

Environment

• Claude Code in VS Code extension
• Long session with a large working file (~2,500-line markdown

document) loaded in context

• Model: Opus 4.7 (1M context)