CUStats is a usage monitoring application designed to track Claude and OpenAI Codex consumption across multiple platforms. The primary purpose is to provide real-time visibility into AI usage limits, specifically Claude's 5-hour and 7-day windows plus OpenAI Codex usage, helping users manage their AI interactions without unexpected interruptions.

The application offers multi-account support for managing both Claude and OpenAI Codex accounts simultaneously. It features pace prediction with weekly projections and color-coded indicators (Healthy/Risky/Over) to forecast consumption. Real-time monitoring displays usage with color-coded progress bars that update instantly. Historical views provide Day, Week, and Month perspectives with GitHub-style activity charts to visualize usage patterns. Session insights include peak usage, average consumption, and detailed statistics for each session window.

The application operates by securely storing user credentials locally with AES-GCM encryption and App Sandbox enabled on macOS. It uses session tokens for Claude and OAuth authentication for OpenAI Codex to fetch usage data without storing sensitive information externally. The system provides real-time status updates through menubar indicators on macOS and mobile notifications on iOS/Android.

Benefits include preventing unexpected limit hits through proactive monitoring and notifications. Use cases involve developers and AI enthusiasts tracking their Claude and OpenAI Codex consumption across personal and work accounts. The privacy-first approach ensures all data remains on-device with zero telemetry or cloud synchronization.

The target audience includes developers, AI researchers, product managers, and remote engineers who regularly use Claude and OpenAI Codex. The application integrates with Claude via session tokens and OpenAI Codex through OAuth, supporting macOS natively with mobile versions for iOS and Android. Technical details highlight local encryption, sandbox security, and platform-specific implementations.