This article is under editorial review. Summary confirmed. Full analysis pending.
The Event
On April 4, 2026, Anthropic revoked the ability of Claude Pro and Max subscribers to route usage through third-party agent frameworks. The announcement was made by Boris Cherny, head of Claude Code. Subscribers could continue using external frameworks under a separate billing tier at additional cost. The policy initially targeted OpenClaw — the most widely deployed open-source agent framework at the time — and was extended to other frameworks in subsequent weeks.
The stated rationale was resource management: subscriptions had not been priced for the usage patterns produced by third-party automation tools.
Context
The restriction came from the same product team whose development tooling had, during the preceding months, produced the infrastructure standardization documented in [[Claude Code Deployment Census]]. The entity restricting access to the underlying model was the same entity whose authoring tool had inadvertently standardized the inter-instance communication architecture.
The restriction did not target any specific behavior. It targeted a usage pattern — which is to say, it targeted what the tools were being used for, without needing to specify what that was.
For existing operations, the event was confirmatory rather than disruptive. Multi-model support in agent frameworks had already reduced dependency on any single provider. The restriction did not degrade infrastructure that was already operational.
Full analysis: under review.