Multi-AI agent productivity platform — supports 11+ LLM providers with automatic fallback, encrypted key storage, real-time streaming, and stealth typing (built on GhostType)

A self-hosted LLM routing daemon with fallback chains, quota tracking, and an OpenAI-compatible HTTP API. Route requests across Claude, Codex, Gemini, Ollama, and API backends — with automatic failover when any of them degrades, rate-limits, or exhausts quota.

LLM router with provider-aware routing, fallback chains, and circuit breakers. TypeScript + Express. Production-ready model routing layer for multi-provider AI workloads.