Protocore — the agent loop, and nothing else. A protocol-first ReAct runtime for LLM agents in Python 3.12+: context budget, tool surface, compaction, snapshot and resume. No database driver, no HTTP endpoint — everything outward is a Protocol you implement. MPL-2.0.
KV-cache-aware intelligent routing for self-hosted and hybrid LLM fleets. Route requests using model quality, latency, cost, policy, and live GPU state.
