Anthropic

Retool connects to Anthropic through secure API credentials stored as encrypted Retool resources. Once configured, teams can invoke Claude models directly within Retool apps and workflows for text generation, analysis, and automation.

Teams explicitly define which queries, fields, and inputs are included in Anthropic prompts inside Retool. Retool never sends data implicitly—only the data you configure in a query or workflow step is transmitted to the model.

Yes. Retool's role-based access controls allow admins to limit which users or groups can create, edit, or run Anthropic-powered queries and workflows. Permissions can be enforced at the app, resource, or environment level.

Anthropic API keys are stored as encrypted Retool resources and are never exposed to end users. Access to those resources is controlled through Retool's permission model.

Yes. Retool supports environment-specific configurations, allowing teams to enable, restrict, or disable Anthropic access in development, staging, or production environments independently.