Worked on the trycua/cua repository to enhance user-facing documentation, focusing on improving clarity around AI model integration and configuration. Addressed onboarding friction by updating Markdown-based guides to clarify the use of the optional ANTHROPIC_API_KEY in post-event export workflows and by providing clearer references to Claude model versions. Added composed-model examples to help users understand grounding and planning features, and standardized terminology by renaming 'VM' to 'sandbox' throughout the documentation. This technical writing effort aimed to reduce misconfiguration risk and support queries, resulting in faster onboarding and better alignment between user expectations and the product’s actual capabilities.