agents.reasoning_and_critique

Reasoning And Critique - Module for reasoning and self-critique agents.

This module provides agents that can reason about their outputs and perform self-critique and reflection to improve their responses.

Available Agents:

  • MCTSAgent: Monte Carlo Tree Search based reasoning agent

Example

Basic MCTS usage:

from haive.agents.reasoning_and_critique.mcts import MCTSAgent, MCTSAgentConfig

config = MCTSAgentConfig(
    name="mcts_reasoner",
    max_iterations=10
)
agent = MCTSAgent(config=config)

result = await agent.ainvoke({"problem": "Solve this logic puzzle..."})

Submodules