Main entry point for complex queries. Returns a woven response after processing through the Nine Worlds.
Retrieves relevant memories from Muninn’s tree.
Stores a new MemoryNode. Returns its ID.
Analyzes query and returns which realms should process it. Returns a dict with primary_realm, secondary_realms, task_type, priority, parallel_safe.
Executes a task graph. Plan is a list of TaskNode objects with dependencies.
- All AI calls go through the
LLMQueue– no concurrent LLM requests. - Realms are stateless; all persistent data lives in Muninn/Helheim.