agents.multi.enhanced_parallel_agent¶

Enhanced ParallelAgent implementation using Agent[AugLLMConfig].

ParallelAgent = Agent[AugLLMConfig] + parallel execution of agents.

Classes¶

`MockExpert`	Init .
`ParallelAgent`	Enhanced ParallelAgent that executes agents concurrently.

Module Contents¶

class agents.multi.enhanced_parallel_agent.MockExpert(name, specialty)¶

Init .

Parameters:

name (str) – [TODO: Add description]
specialty (str) – [TODO: Add description]

async arun(input_data)¶

Arun.

Parameters:: input_data (str) – [TODO: Add description]
Returns:: Add return description]
Return type:: [TODO

class agents.multi.enhanced_parallel_agent.ParallelAgent¶

Bases: haive.agents.simple.enhanced_simple_real.EnhancedAgentBase

Enhanced ParallelAgent that executes agents concurrently.

ParallelAgent = Agent[AugLLMConfig] + parallel execution + result aggregation.

All agents receive the same input and execute concurrently. Results can be aggregated using various strategies (all, first, best, majority).

agents¶: List of agents to execute in parallel

aggregation_strategy¶: How to combine results

timeout_per_agent¶: Timeout for each agent

min_agents_for_consensus¶: Minimum agents for consensus strategies

Examples

Parallel analysis with multiple experts:

experts = ParallelAgent(
    name="expert_panel",
    agents=[
        FinanceExpert(),
        TechExpert(),
        MarketExpert()
    ],
    aggregation_strategy="all"
)

# All experts analyze simultaneously
results = experts.run("Analyze startup investment opportunity")

Consensus-based decision making:

validators = ParallelAgent(
    name="validation_ensemble",
    agents=[Validator1(), Validator2(), Validator3()],
    aggregation_strategy="majority",
    min_agents_for_consensus=2
)

# Returns majority consensus
decision = validators.run("Is this transaction valid?")

add_agent(agent)¶

Add an agent to the parallel group.

Parameters:: agent (haive.agents.simple.enhanced_simple_real.EnhancedAgentBase) – Agent to add
Return type:: None

build_graph()¶

Build parallel execution graph.

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

async execute_parallel(input_data)¶

Execute all agents in parallel.

Parameters:: input_data (Any) – Input for all agents
Returns:: Aggregated results based on strategy
Return type:: list[Any] | Any

remove_agent(agent)¶

Remove an agent from the parallel group.

Parameters:: agent (haive.agents.simple.enhanced_simple_real.EnhancedAgentBase) – Agent to remove
Returns:: True if removed, False if not found
Return type:: bool

setup_agent()¶

Setup parallel coordinator.

Return type:: None

classmethod validate_agents(v)¶

Validate agent list.

Parameters:: v (list[haive.agents.simple.enhanced_simple_real.EnhancedAgentBase])
Return type:: list[haive.agents.simple.enhanced_simple_real.EnhancedAgentBase]