Reproduced behavior shows Claude Code disconnects sessions and burns usage limits when git commit messages or chat content references “OpenClaw.”
Key Takeaways
Keyword matching on “OpenClaw” in commit messages triggers immediate session disconnection and drives usage to 100% of limit.
Mentioning openclaw.ai via direct link in chat also ends sessions and hits 5-hour usage caps, not just agentic/commit contexts.
The behavior appears to be a targeted suppression of OpenClaw-related workloads, likely tied to capacity pressure from that tool’s user base.
Claude Code has a pattern of shipped bugs (HERMES.md, thinking-message pruning, cache-skipping) consistent with a vibe-coded codebase.
Hacker News Comment Review
Commenters broadly reproduced the issue: a single git commit -m containing an OpenClaw schema string is enough to trigger the penalty, making it deterministic and not a fluke.
The dominant inference is that Anthropic views OpenClaw-driven load as an existential capacity threat, and this keyword block is a reactive and heavy-handed mitigation rather than a policy with legitimate UX rationale.
Concern extends past censorship to code quality: commenters note Claude Code is reportedly vibe-coded by its own team, and the string of recent bugs makes that credible.
Notable Comments
@abdullin: Provided a minimal five-command repro using a bare OpenClaw JSON schema in a commit message, confirming 100% usage spike.
@trb: “Claude Code lead has said many times that he/his team don’t read the code” – links vibe-coding culture to this and other recent Claude Code bugs.
@threecheese: Points to a public “AI Hype” board filled with subscription-bypass toolkits as context for why Anthropic may be acting aggressively against certain traffic patterns.
