False-positive Usage Policy block on benign procedural-animation prompt
Request ID: req_011Cbxar68EFGgqq16j36k2v
Date: 2026-06-11
Surface: Claude Code desktop app (Windows 11)
What happened
A turn was hard-blocked with "Request was blocked - Claude Code is unable to respond to this request, which appears to violate our Usage Policy."
The prompt was entirely benign game-development research:
"Research best practices for procedural walk-cycle animation - gait biomechanics, weight shift, pelvic rotation, foot roll, arm follow-through - and distill it into a playbook mapped to our procPose/LOCO_PARAMS knobs."
The block fired mid-turn while the assistant was reading project source files (a procedural character-animation module for a zombie survival game).
Why this is a false positive
The request is standard game-animation engineering (gait biomechanics terminology). The identical request retried in a fresh turn moments later was processed without issue, and the project has been doing the same procedural-rigging work for weeks without blocks. Likely an unlucky token combination in the full context (the game codebase contains words like "wound", "stab", "knife") tripping the safety classifier.
Expected
Benign biomechanics/animation research prompts should not be blocked.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗