Llama-Conductor Provides Router And Grounded Memory Harness

Llama-conductor is an open-source Python router and RAG harness that integrates llama-swap, llama.cpp, Qdrant, and popular frontends to provide deterministic memory and grounded reasoning. It implements Vodka (CTC context trimming and Total Recall), Mentats for Vault-based verification, and filesystem KB integration to reduce hallucinations, goldfish memory, and context bloat. Install via pip and run the router, then connect model runners, llama-swap, and Qdrant for Vault features.