Anthropic has blocked Claude Code subscription holders from using OpenClaw, an autonomous AI agent that the company characterized as a heavy consumer of API resources. The restriction surfaced in discussions on Hacker News, where users reported being locked out of workflows that pipe development tasks through Claude's infrastructure.
OpenClaw runs as an autonomous agent capable of executing multi-step tasks without human hand-holding—testing code, iterating on fixes, and managing development pipelines. That autonomy burns through API calls at a rate that dwarfs typical usage. One Hacker News commenter described such tools as "walking attack surfaces" for their unpredictable, intensive consumption.
Theories about Anthropic's motivation split three ways. Capacity constraints: autonomous agents stress infrastructure in ways individual users don't. Economics: Claude Code subscriptions weren't priced for users running continuous automated loops. Control: Anthropic may prefer to keep its agent ecosystem closed rather than enable third-party frameworks like AGENTS.MD.
Anthropic hasn't issued a public statement. Developers who built workflows around OpenClaw are now reassessing whether subscription-based API access can support serious agent development—or if the economics only work for human-paced interaction.