Claude subscriptions now eligible for monthly Agent SDK credits

Anthropic has changed how paid Claude subscriptions interact with its agent tooling. In a Help Center article published May 19, the company said that starting June 15, 2026, usage from Claude Agent SDK and the `claude -p` command will no longer count against a subscriber’s regular Claude plan limits. Instead, eligible users on Pro, Max, Team and Enterprise plans can request a separate monthly credit for that usage. The change matters because Anthropic is drawing a clearer line between interactive Claude use and agent-driven workloads. The Help Center article says normal subscription limits will remain reserved for “interactive use” of Claude Code, Claude Cowork and Claude, while the new credit applies only to Agent SDK usage and resets with the billing cycle. Anthropic also says Claude Platform accounts using API keys do not get the credit and that pay-as-you-go billing remains in place. (support.claude.com) ### What exactly is Anthropic changing on June 15? Anthropic’s May 19 support document says the main change takes effect on June 15, 2026. From that date, Agent SDK activity and usage through `claude -p` stop consuming the same usage bucket as a paid Claude subscription’s interactive sessions. (support.claude.com) The Help Center article describes the new benefit as a “separate monthly credit” for eligible subscribers. That credit is tied to Agent SDK usage only, not to the broader Claude API, and it renews on the user’s billing cycle, according to the documentation. ### Which Claude customers get the credit, and who does not? (support.claude.com) Anthropic says the change is available on Pro, Max, Team and Enterprise plans. The company’s support page says those subscribers can request the monthly credit once the program begins on June 15. Claude Platform customers using API keys are excluded. (support.claude.com) Anthropic says in the same article that API-key users will continue under pay-as-you-go billing, which means the new subscription-linked credit is aimed at Claude plan subscribers rather than developers buying usage directly through the platform. ### What usage does the credit actually cover? Anthropic says the credit applies to three buckets: Agent SDK features, the `claude -p` command, and third-party apps built on the Agent SDK. That makes the change broader than a narrow developer-only perk, because Anthropic’s own documentation explicitly includes outside applications that run on the SDK. (support.claude.com) The support article does not say that ordinary Claude API calls are folded into the credit. Instead, it distinguishes Agent SDK usage from API-key billing, which remains separate. ### How does Cloudflare fit into this story? Cloudflare said on May 19 that it integrated with Anthropic’s Claude Managed Agents to provide what it called a “fast, isolated execution environment for autonomous code delivery.” In its announcement, Cloudflare said the setup lets builders scale agent workflows while controlling access to private backends and customizing tools and runtimes. (support.claude.com) Cloudflare’s post also said the integration adds sandbox controls, observability, private service connectivity, browser controls, email support and custom tools. The company described the product as a way to run Claude Managed Agents with more infrastructure control than Anthropic-hosted defaults alone. ### What should users watch next? (blog.cloudflare.com) June 15, 2026 is the next concrete date in Anthropic’s rollout. That is when Pro, Max, Team and Enterprise subscribers become eligible for the separate Agent SDK credit, according to Anthropic’s Help Center. Cloudflare’s May 19 post points developers to its Claude Managed Agents integration as the infrastructure layer for teams that want isolated agent execution environments. (blog.cloudflare.com) Anthropic’s support page is the source for plan eligibility and credit rules, while Cloudflare’s announcement lays out the sandbox features available through the integration. (support.claude.com)