haive.core.graph.node.multi_agent_node¶
Multi-agent node with hierarchical state projection.
This module provides node configurations for multi-agent systems that properly handle state projection between the container state and individual agent states.
Classes¶
Node for executing agents within a multi-agent state container. |
|
Generic state projection node for any schema transformation. |
Functions¶
|
Create a multi-agent node for executing an agent from MultiAgentState. |
|
Create a state projection node. |
Module Contents¶
- class haive.core.graph.node.multi_agent_node.MultiAgentNode¶
Bases:
haive.core.graph.node.base_node_config.BaseNodeConfig[haive.core.schema.prebuilt.multi_agent_state.MultiAgentState,haive.core.schema.prebuilt.multi_agent_state.MultiAgentState]Node for executing agents within a multi-agent state container.
This node handles: - State projection from MultiAgentState to agent-specific schema - Agent execution with its expected state type - State updates back to the container - Recompilation tracking
The key innovation is that each agent receives its exact expected state schema, not a flattened global state.
- validate_config()¶
Validate node configuration.
- Return type:
Self
- class haive.core.graph.node.multi_agent_node.StateProjectionNode¶
Bases:
haive.core.graph.node.base_node_config.BaseNodeConfig[TInput,TOutput]Generic state projection node for any schema transformation.
This node can project from any input schema to any output schema, useful for bridging between different state representations.
- haive.core.graph.node.multi_agent_node.create_multi_agent_node(agent_name, name=None, **kwargs)¶
Create a multi-agent node for executing an agent from MultiAgentState.
- Parameters:
- Return type:
- haive.core.graph.node.multi_agent_node.create_projection_node(input_schema, output_schema, name=None, **kwargs)¶
Create a state projection node.
- Parameters:
- Return type: