xTokens

Universal multi-provider AI token metering, rolling rate-limit forecasting, and advisory budget locks for macOS.

1. Executive Summary

Autonomous coding swarms and multi-agent development pipelines consume tokens at unprecedented rates. Without cross-provider telemetry, teams face unexpected budget overruns, abrupt rate-limit throttles, and degraded agent responsiveness.

xTokens is a native macOS and CLI observability engine built to monitor, forecast, and govern token consumption across every major AI provider. Built in Swift and C with universal Apple Silicon and Intel binaries, xTokens intercepts provider usage streams with sub-millisecond precision, providing calibrated rolling usage windows, predictive burn rate alerts, and non-blocking advisory locks.

2. Core Capabilities

• Unified Multi-Provider Ingestion: Calibrated adapters for OpenAI (ChatGPT), Anthropic (Claude), Google Gemini, xAI (Grok), OpenRouter (Hermes), Cursor, and local Ollama runtimes.
• Rolling Window Forensics: Track 5-hour burst limits, 7-day weekly rolling windows, and monthly billing caps simultaneously.
• Proactive Anti-Stall Advice: Provides immediate proceed, conserve, or hold signals directly to autonomous agent dispatchers before rate limits trigger.
• Privacy First: Meters tokens and costs without capturing or storing prompts, completions, or API keys.
• Developer ID & Hardened Runtime: Signed with Apple Developer ID ZU2882L4HT and fully notarized for enterprise fleet deployment.