Anthropic bans OAuth tokens (including Agent SDK) in 3P tools (code.claude.com)

Anthropic has implemented a ban on the use of OAuth tokens, including those from the Agent SDK, in third-party tools. This decision affects developers who have been using OAuth tokens obtained via the Free, Pro, and Max plans to integrate access to Claude’s capabilities in their applications. Instead, developers are now required to utilize API key authentication through Claude Console or other supported cloud providers to maintain compliance with the updated terms of service. This move is significant for the AI/ML community as it emphasizes a stricter control over authentication methods, reinforcing the need for secure and compliant integration practices. By mandating API keys for third-party interactions, Anthropic aims to enhance security measures and trust in its AI services, ensuring that user data is handled in accordance with established legal agreements. The shift not only safeguards user credentials but also aligns with best practices around data retention and healthcare compliance, particularly for customers operating under Business Associate Agreements (BAA).