Files
agents_core/agents_core
sylvain 4b5960ada6 feat: global LLM coordinator to prevent Ollama overload
- llm_coordinator.py: LLMCoordinator server (runs in Nexus)
  FIFO queue + semaphore, max_concurrent configurable
  local_acquire/release for Nexus, MQTT request/release for sub-agents
- base_agent.py: _llm_slot_acquire/_llm_slot_release around _llm_loop
  MQTT round-trip to agents/llm/request, 90s timeout degraded fallback
  Enabled via use_llm_coordinator: true in config

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-15 19:32:27 +00:00
..