agents.multi.enhanced_multi_agent_generic¶

Enhanced MultiAgent with proper generics for contained agents.

MultiAgent[AgentsT] where AgentsT represents the agents it contains.

Classes¶

`AdaptiveBranchingMultiAgent`	Branching MultiAgent that adapts routing based on performance.
`BranchingMultiAgent`	MultiAgent specialized for branching execution.
`ConditionalMultiAgent`	MultiAgent with conditional execution based on previous results.
`MultiAgent`	Enhanced MultiAgent generic on the agents it contains.
`ReportTeamAgents`	Typed dict for report team agents.

Module Contents¶

class agents.multi.enhanced_multi_agent_generic.AdaptiveBranchingMultiAgent¶

Bases: BranchingMultiAgent

Branching MultiAgent that adapts routing based on performance.

Tracks agent performance and adjusts routing probabilities.

get_best_agent_for_task(task_type)¶

Get best performing agent for task type.

Parameters:: task_type (str)
Return type:: str

update_performance(agent_name, success, duration)¶

Update agent performance metrics.

Parameters:

agent_name (str)
success (bool)
duration (float)

Return type:

None

class agents.multi.enhanced_multi_agent_generic.BranchingMultiAgent¶

Bases: MultiAgent[dict[str, Agent]]

MultiAgent specialized for branching execution.

Routes to different agents based on conditions.

build_graph()¶

Build branching execution graph.

Return type:: haive.core.graph.state_graph.base_graph2.BaseGraph

class agents.multi.enhanced_multi_agent_generic.ConditionalMultiAgent¶

Bases: MultiAgent[dict[str, Agent]]

MultiAgent with conditional execution based on previous results.

Executes agents conditionally based on outputs.

should_continue(state, current_agent)¶

Determine next agent based on conditions.

Parameters:

state (dict[str, Any])
current_agent (str)

Return type:

str | None

class agents.multi.enhanced_multi_agent_generic.MultiAgent¶

Bases: Agent, Generic[AgentsT]

Enhanced MultiAgent generic on the agents it contains.

MultiAgent[AgentsT] = Agent[AugLLMConfig] + agents: AgentsT

This properly represents that MultiAgent is: 1. An agent itself (uses AugLLMConfig for coordination) 2. Generic on the agents it contains

Examples

With typed dict of agents:

agents: Dict[str, Agent] = {
    "planner": PlannerAgent(...),
    "executor": ExecutorAgent(...)
}
multi: MultiAgent[Dict[str, Agent]] = MultiAgent(
    name="coordinator",
    agents=agents
)

With list of agents:

agent_list: List[ReactAgent] = [agent1, agent2, agent3]
multi: MultiAgent[List[ReactAgent]] = MultiAgent(
    name="ensemble",
    agents=agent_list
)

With specific agent types:

from typing import TypedDict

class MyAgents(TypedDict):
    researcher: RAGAgent
    analyzer: ReactAgent
    writer: SimpleAgent

agents = MyAgents(
    researcher=rag_agent,
    analyzer=react_agent,
    writer=simple_agent
)

multi: MultiAgent[MyAgents] = MultiAgent(
    name="report_team",
    agents=agents
)

get_agent(name)¶

Get agent by name.

Parameters:: name (str)
Return type:: Agent | None

get_agent_names()¶

Get list of agent names.

Return type:: list[str]

classmethod validate_agents(v)¶

Validate agents based on type.

Parameters:: v (AgentsT)
Return type:: AgentsT

class agents.multi.enhanced_multi_agent_generic.ReportTeamAgents¶

Bases: TypedDict

Typed dict for report team agents.

Initialize self. See help(type(self)) for accurate signature.