> ## Documentation Index
> Fetch the complete documentation index at: https://pathlit.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent Swarm Node

> Orchestrate multiple AI agents to collaborate and solve complex tasks

## Overview

The Agent Swarm Node enables multiple AI agents to work together collaboratively to solve complex tasks. This node allows you to:

* Configure multiple specialized agents with different capabilities
* Define an initial agent to start the conversation
* Allow agents to hand off tasks to other agents based on their expertise
* Create complex workflows where agents collaborate to achieve a goal

## Configuration Parameters

The Agent Swarm Node requires configuration of multiple agents that will work together in the swarm. Each agent can be configured independently with its own settings and capabilities.

### Node Configuration

* **Query**:

  Specify the task or question for the swarm to handle. For example,

  > *Given this email: <Tooltip tip="email id from the gmail receive node">\{email id}</Tooltip>, identify if it is asking for availability. Look at my calendar to check for my availability and provide timeslots to respond to the email.*

  > *Given a contact, conduct web search on the person, and then draft a personalized outreach email to ...*

* **Agents**

  Configure multiple agents that will participate in the swarm. For each agent, you can:

  * Set a custom display name
  * Designate it as the initial agent
  * Configure agent-specific settings and tools
  * Define the agent's role and capabilities

  <Tip>Choose the initial agent carefully as it will be responsible for starting the conversation and delegating tasks to other agents.</Tip>

<AccordionGroup>
  <Accordion title="Advanced Settings">
    <ul>
      <li>
        <strong>Max Steps</strong>: Set the maximum number of steps the swarm can take (1-200). This prevents infinite loops and ensures the swarm reaches a conclusion.
      </li>
    </ul>
  </Accordion>
</AccordionGroup>

## Expected Inputs and Outputs

* **Inputs**:

  * The node accepts input data that can be referenced in the query using format strings.

* **Outputs**:

  * **output**: The final result produced by the swarm
  * **conversation**: A detailed log of the conversation between agents, showing how they collaborated to reach the solution

## Use Case Examples

1. **Complex Problem Solving**:

   Configure multiple agents with different expertise to break down and solve complex problems. Each agent can handle specific aspects of the problem based on their specialized knowledge.

2. **Multi-perspective Analysis**:

   Use different agents to analyze the same problem from various angles, providing a comprehensive understanding of the situation.

3. **Collaborative Content Creation**:

   Set up agents with different writing styles and expertise to collaboratively create content, with each agent contributing their specialized knowledge.

## Error Handling and Troubleshooting

* **Agent Configuration**:

  Make sure each agent in the swarm is properly configured with appropriate settings and tools for their intended role.

* **Max Steps Exceeded**:

  If the swarm is not reaching a conclusion, you may need to increase the max\_steps parameter or optimize the agent configurations.

If you encounter issues not covered in this documentation, please contact our [support team](mailto:dev@pathlit.ai) for assistance.

## Relevant Nodes

<CardGroup>
  <Card title="Agent Node" href="/docs/nodes/agent">
    Generic Agent Node
  </Card>

  <Card title="Gmail Agent Node" href="/docs/nodes/gmail-agent">
    Agent that interact with Gmail
  </Card>

  <Card title="Google Calendar Agent Node" href="/docs/nodes/gcal-agent">
    Agent that interact with Google Calendar
  </Card>

  <Card title="Bitbucket Agent Node" href="/docs/nodes/bitbucket-agent">
    Agent that interact with Bitbucket
  </Card>
</CardGroup>