Something strange is happening with Claude Code when users mention OpenClaw. Anthropic's coding assistant either refuses requests, ends sessions without warning, or burns through your usage limits when commits or messages reference OpenClaw, an open-source personal AI assistant built by developer @steipete.
One user showed that a git commit containing "openclaw.inbound_meta.v1" caused an instant disconnection. Session usage spiked to 100%. Another developer pasted a direct link to openclaw.ai and their chat died on the spot, triggering a 5-hour usage limit despite barely using Claude that day. A third said Claude asked if "OpenClaw" was a typo, then killed the session.
OpenClaw runs locally instead of in a cloud walled garden. It handles email, calendar coordination, flight check-ins, and autonomous tasks through WhatsApp and Telegram. People call it "magic" and say it's a "teammate rather than a tool" thanks to persistent memory and multi-agent management.
ChatGPT, Gemini, and Copilot nudge users toward their own products. We've seen that. But killing sessions over a competitor's name is something else. If Claude Code is actively detecting and punishing mentions of a specific rival, we're in new territory. A tool that locks you out for knowing about the competition.
Anthropic hasn't commented publicly.