Just as developers were still digesting the Claude Code-induced identity crisis, Anthropic changed the rules. The company blocked third-party tools that were spoofing Claude Code to get cheap access to Opus 4.5. Moreover, Anthropic has now also clearly mentioned that users cannot use Claude to develop competing platforms. Thariq Shihipar, who works on Claude Code at Anthropic, said they had "tightened safeguards against spoofing the Claude Code harness." The wrappers are simply not coming back. To make it clear, tools like OpenCode had been letting people connect their $200 Claude subscription to automated agent loops. By pretending to be the official Claude Code client, they could run massive workloads without paying API prices. These third-party tools effectively removed the speed limit Anthropic had on its $200 plan. Some users burned through what would have cost over $1,000 a month on the API. Anthropic banned it. Shortly after, OpenCode reaffirmed that users can access ChatGPT Plus and Pro plans in OpenCode, giving access to Codex as well. At the same time, Anthropic cut off xAI. Elon Musk's lab had been using Claude through Cursor to help build its own models—a clear violation of Anthropic's terms. This was not a one-off. Anthropic had cut off OpenAI in August 2025. Windsurf was blocked in June, forcing it to move to a bring-your-own-key model. It's now clear that Claude is open until it threatens Anthropic's moat. Much of Claude Code's popularity came from people not really using Claude Code at all, but spoofing it to run Opus 4.5 wherever they wanted. That era now seems to be ending almost as soon as it started. |
Комментариев нет:
Отправить комментарий
Примечание. Отправлять комментарии могут только участники этого блога.