haive.agents.research.open_perplexity.config

Configuration for the open_perplexity research agent.

from typing import Any This module defines the configuration class and factory methods for creating research agent configurations. It includes settings for LLM engines, tools, vector stores, and research parameters.

Classes

ResearchAgentConfig

Configuration for open_perplexity research agent.

Module Contents

class haive.agents.research.open_perplexity.config.ResearchAgentConfig

Bases: haive.core.engine.agent.agent.AgentConfig

Configuration for open_perplexity research agent.

Defines all configuration parameters for the research agent, including state schemas, engines, tools, and research parameters.

state_schema

Schema for the agent state

input_schema

Schema for input to the agent

output_schema

Schema for agent output

engines

Dictionary of AugLLM engines for different tasks

tools

Tools for research and analysis

vectorstore_config

Vector store configuration for document storage

react_agent_name

Name of the configured ReAct agent

rag_agent_name

Name of the configured RAG agent

report_format

Format for the final report

research_depth

Depth of research (1-5, higher means more thorough)

max_sources_per_query

Maximum number of sources to use per query

concurrent_searches

Number of concurrent searches to perform

default_report_sections

Default sections for the research report

classmethod from_scratch(name=None, llm_model='gpt-4o', **kwargs)

Create a research agent configuration from scratch.

Factory method to create a fully configured research agent with all necessary engines, tools, and settings.

Parameters:

• name (str | None) – Optional name for the agent (defaults to timestamped name)

• llm_model (str) – Model to use for the agent (default: “gpt-4o”)

• **kwargs – Additional configuration parameters

Returns:

A fully configured research agent configuration

Return type:

ResearchAgentConfig