Dev hitting major friction with Claude API: over 50% of requests are getting blocked while maintaining PPQdotAI. Not doing anything sketchy, just standard website maintenance tasks, but Anthropic's safety filters are triggering constantly. Either there's overly aggressive content moderation happening or the dev got flagged somehow. This is a real problem for anyone building production apps on Claude - unreliable API access kills developer velocity. If you're using Claude in your stack, you might want to have a fallback model ready. The inconsistency here is brutal for any serious dev workflow.