Chain Agent

The ChainAgent is an agent class in the Agent Squad System that allows for the sequential execution of multiple agents. It processes a request by passing the output of one agent as input to the next, creating a chain of agent interactions.

Creating a ChainAgent

Basic Example

Python Package

If you haven’t already installed the AWS-related dependencies, make sure to install them:

pip install "agent-squad[aws]"

Here’s how to create a ChainAgent with only the required parameters:

TypeScript
Python

import { ChainAgent, ChainAgentOptions } from 'agent-squad';
import { BedrockLLMAgent } from 'agent-squad';

const agent1 = new BedrockLLMAgent({
  name: 'Agent 1',
  description: '..AGENT DESCRIPTION..'
});

const agent2 = new BedrockLLMAgent({
  name: 'Agent 2',
  description: '..AGENT DESCRIPTION..'
});

const chainAgent = new ChainAgent({
  name: 'Chain Tech Agent',
  description: 'Specializes in technology areas including software development, hardware, AI, cybersecurity, blockchain, cloud computing, emerging tech innovations, and pricing/costs related to technology products and services.',
  agents: [agent1, agent2]
});

from agent_squad.agents import ChainAgent, ChainAgentOptions
from agent_squad.agents import BedrockLLMAgent, BedrockLLMAgentOptions

agent1 = BedrockLLMAgent(BedrockLLMAgentOptions(
    name='Agent 1',
    description='..AGENT DESCRIPTION..'
))

agent2 = BedrockLLMAgent(BedrockLLMAgentOptions(
    name='Agent 2',
    description='..AGENT DESCRIPTION..'
))

chain_agent = ChainAgent(ChainAgentOptions(
    name='BasicChainAgent',
    description='A simple chain of multiple agents',
    agents=[agent1, agent2]
))

Intermediate Example

This example shows how to create a ChainAgent with a custom default output:

TypeScript
Python

import { ChainAgent, ChainAgentOptions } from 'agent-squad';
import { BedrockLLMAgent } from 'agent-squad';

const agent1 = new BedrockLLMAgent({
  name: 'Agent 1',
  description: '..AGENT DESCRIPTION..'
});

const agent2 = new BedrockLLMAgent({
  name: 'Agent 2',
  description: '..AGENT DESCRIPTION..',
  streaming: true
});

const chainAgent = new ChainAgent({
  name: 'IntermediateChainAgent',
  description: 'A chain of agents with custom default output',
  agents: [agent1, agent2],
  defaultOutput: 'The chain encountered an issue during processing.'
});

from agent_squad.agents import ChainAgent, ChainAgentOptions
from agent_squad.agents import BedrockLLMAgent, BedrockLLMAgentOptions

agent1 = BedrockLLMAgent(BedrockLLMAgentOptions(
    name='Agent 1',
    description='..AGENT DESCRIPTION..'
))

agent2 = BedrockLLMAgent(BedrockLLMAgentOptions(
    name='Agent 2',
    description='..AGENT DESCRIPTION..'
))

chain_agent = ChainAgent(ChainAgentOptions(
    name='IntermediateChainAgent',
    description='A chain of agents with custom default output',
    agents=[agent1, agent2],
    default_output='The chain encountered an issue during processing.'
))

Advanced Example

For more complex use cases, you can create a ChainAgent with all available options:

TypeScript
Python

import { ChainAgent, ChainAgentOptions } from 'agent-squad';
import { BedrockLLMAgent } from 'agent-squad';

const agent1 = new BedrockLLMAgent({
  name: 'Agent 1',
  description: '..AGENT DESCRIPTION..'
});

const agent2 = new BedrockLLMAgent({
  name: 'Agent 2',
  description: '..AGENT DESCRIPTION..',
  streaming: true
});

const options: ChainAgentOptions = {
  name: 'AdvancedChainAgent',
  description: 'A sophisticated chain of agents with all options',
  agents: [agent1, agent2],
  defaultOutput: 'The chain processing encountered an issue.',
  saveChat: true
};

const chainAgent = new ChainAgent(options);

from agent_squad.agents import ChainAgent, ChainAgentOptions
from agent_squad.agents import BedrockLLMAgent, BedrockLLMAgentOptions

agent1 = BedrockLLMAgent(BedrockLLMAgentOptions(
    name='Agent 1',
    description='..AGENT DESCRIPTION..'
))

agent2 = BedrockLLMAgent(BedrockLLMAgentOptions(
    name='Agent 2',
    description='..AGENT DESCRIPTION..',
    streaming=True
))

options = ChainAgentOptions(
    name='AdvancedChainAgent',
    description='A sophisticated chain of agents with all options',
    agents=[agent1, agent2],
    default_output='The chain processing encountered an issue.',
    save_chat=True
)

chain_agent = ChainAgent(options)

Integrating ChainAgent into the Agent Squad

To integrate the ChainAgent into your Agent Squad:

TypeScript
Python

import { AgentSquad } from "agent-squad";

const orchestrator = new AgentSquad();
orchestrator.addAgent(chainAgent);

from agent_squad.orchestrator import AgentSquad

orchestrator = AgentSquad()
orchestrator.add_agent(chain_agent)

Streaming Responses

The ChainAgent supports streaming responses only for the last agent in the chain.

This design ensures efficient processing through the chain while still enabling streaming capabilities for the end result.

By leveraging the ChainAgent, you can create sophisticated, multi-step processing pipelines within your Agent Squad system, allowing for complex interactions and transformations of user inputs, with the added flexibility of streaming output from the final processing step.