agents

Haive Agents Module - Main exports.

This module provides various agent implementations for the Haive framework, including base agents, multi-agents, reactive agents, and specialized agents for different use cases like RAG, planning, memory, and reasoning.

Key Components:

  • Base: Core agent abstractions and foundational classes

  • Simple: Basic agent implementations for common use cases

  • React: Reactive agents with reasoning loops and tool usage

  • Multi: Multi-agent coordination and orchestration

  • RAG: Retrieval-Augmented Generation agents

  • Planning: Planning and execution agents

  • Memory: Memory-enabled agents with long-term context

  • Conversation: Conversational agent patterns

  • Supervisor: Agent supervision and coordination patterns

The agents are designed to be modular, extensible, and optimized for various AI workflows and use cases within the Haive ecosystem.

Examples

Basic agent usage:

from haive.agents import SimpleAgent
from haive.core.engine.aug_llm import AugLLMConfig

agent = SimpleAgent(
    name="helper",
    engine=AugLLMConfig(model="gpt-4")
)
result = agent.run("Hello world")

Multi-agent coordination:

from haive.agents import MultiAgent, SimpleAgent, ReactAgent

coordinator = MultiAgent([
    SimpleAgent(name="planner"),
    ReactAgent(name="executor", tools=[...])
], mode="sequential")

RAG agent setup:

from haive.agents.rag import BaseRAGAgent
from haive.core.models import VectorStoreConfig

rag_agent = BaseRAGAgent(
    vectorstore_config=VectorStoreConfig(...)
)

Submodules