agents.reasoning_and_critique.self_discover.selector.agent¶

Self-Discover Selector Agent implementation.

Classes¶

SelectorAgent

Agent that selects relevant reasoning modules for a given task.

Module Contents¶

class agents.reasoning_and_critique.self_discover.selector.agent.SelectorAgent(name='selector', engine=None, **kwargs)¶

Bases: haive.agents.simple.SimpleAgent

Agent that selects relevant reasoning modules for a given task.

The Selector Agent is the first stage in the Self-Discover workflow. It analyzes the task and selects 3-5 reasoning modules from the available options that would be most effective for solving the problem.

name¶: Agent identifier (default: “selector”)

engine¶: LLM configuration for the agent

Example

>>> from haive.core.engine.aug_llm import AugLLMConfig
>>>
>>> config = AugLLMConfig(temperature=0.3)
>>> selector = SelectorAgent(engine=config)
>>>
>>> result = await selector.arun({
...     "available_modules": "1. Critical thinking\\n2. Pattern recognition...",
...     "task_description": "Design a recommendation system"
... })

Initialize the Selector Agent.

Parameters:

name (str) – Name for the agent
engine (haive.core.engine.aug_llm.AugLLMConfig) – LLM configuration (if not provided, creates default)
**kwargs – Additional arguments passed to SimpleAgent