Claude

Anthropic restricted subscription OAuth tokens to work only with the official Claude Code CLI, blocking tools like OpenCode, Moltbot, and integrations in Cursor. Users who built workflows around third-party tools were locked out mid-project, forcing them to either downgrade subscriptions or abandon the platform entirely.

Figma's platform is effectively closed for programmatic design updates. MCP integration with Claude is read-only, and meaningful mutations require plugins running inside Figma's sandbox. There is no serious external API for automation, CI-style pipelines, or integration with engineering systems.

Claude frequently ignores explicit instructions to make minimal code changes and instead reimplements entire sections from scratch, making it unsuitable for production work where careful control is required.

Heavy usage of Claude Code through $200/month Max subscription (unlimited tokens) would cost $1,000+ monthly via API, creating economic pressure to use subscription workarounds, which Anthropic now blocks, leaving no cost-effective option for autonomous agent automation.

Developers must manually switch between Copilot models (ChatGPT, Claude, etc.) for different task types, often forgetting to optimize cost. There is no smart triage system to route simple questions to cheaper models and complex requests to premium models.

Anthropic retains user chat and coding session data (de-identified) for up to 5 years for model improvement unless explicitly disabled, creating a default opt-in model training practice that applies to new or resumed chats after enabling the setting.

Claude model weights are not open-source, preventing developers from deploying models on-premise or customizing them for specific use cases, unlike competitors like Mistral or LLaMA 3.