tool-routing
The pre-LLM reflex layer for AI agents. library with a live training server, microsecond latency, learns from LLM corrections — System 1 to your LLM's System 2.
Budget-aware context compilation and context firewall for tool-heavy AI agents.
A retrieval-gated skill architecture for LLM agents that scales to hundreds of tools by exposing only the top-K relevant capabilities per request.
Pick the right tools for an agent task. Boring baseline, open dataset (14K traces from public agent benchmarks).
