awrouter
OpenRouter for your own fleet: pick a model backend by cost/latency/ capability, fail over, fit the context window, stream. Standalone, OpenAI-compatible, no Aither-specifics required to be valuable.
The Aither World is an operating system for agents — a Linux you can hand to one, the runtimes it works in, and the tools it works with. awrouter is one of its 71 bricks — each installs on its own, runs offline, and needs no account. 5 more are in development and 15 are planned.
The problem
Every LLM call in the platform routes through MicroScheduler (:8150) -- but the routing logic (backend discovery, failover, tier maps, context fitting, streaming) is buried inside a 19,535-line service with 107 monorepo imports. A stranger cannot run it: it demands the internal CA, services.yaml discovery, secrets at :8111, the fleet's model inventory. The routing plane is valuable standalone and entangled by accident of packaging, not by design.
Adopt it
pip install awrouterPoint it at your backends (vLLM / llama.cpp / any OpenAI-compatible endpoint); ask it which backend should serve model X right now, or to stream a completion with failover and context fitting.
Pairs with
Pairing is composition, never dependency — awrouter installs and runs on its own.
The rest of the Aither World
Every brick below is live, drawn from that repository's own published manifest.