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

# Parent agents overview

> Understand how parent agents orchestrate workflows and where to configure them

## What a parent agent does

Parent agents orchestrate workflows. They run on chat or triggers, call tools, and coordinate sub-agents to complete tasks.

<Frame>
  <img src="https://mintcdn.com/maadify/X4aEKWAIWIr8YAhZ/images/parent-agents/agent_configurations.png?fit=max&auto=format&n=X4aEKWAIWIr8YAhZ&q=85&s=8e50ae72124ec5141f67b8285af8ecf4" alt="Parent agent configuration showing tabs for sub-agents, rules, and triggers" width="1536" height="825" data-path="images/parent-agents/agent_configurations.png" />
</Frame>

## Core building blocks

* **Sub-agents** define the prompts and outputs used in your workflow.
* **Tool executions** run connector tools with templated inputs.
* **Rules** evaluate conditions and route to the next step.
* **Triggers** start the workflow from schedules or webhooks.
* **Approvals and errors** pause or stop execution based on outcomes.

## Create a parent agent

<Steps>
  <Step title="Open the create form">
    Go to **Agent Configurations** and select **Add Parent Agent**.
  </Step>

  <Step title="Choose a Parent Agent Type">
    **Parent Agent Type** is the high-level category for your agent (for example, **General Agent**). It identifies what kind of parent agent you are building and is used to group agents of the same kind.
  </Step>

  <Step title="Choose an Action">
    **Action** is a sub-category of the selected **Parent Agent Type**. It narrows down the agent's primary purpose within its category and helps further classify the parent agent. The available actions depend on the type you selected; if a type has no actions, this field is hidden.
  </Step>

  <Step title="Name the agent">
    Enter a **Parent Agent Name** that is unique and reflects the agent's category and action. The **Parent Agent Name Key** is generated automatically from the name and used as the agent's stable identifier.
  </Step>

  <Step title="Define purpose and scope">
    Add a clear **Description** of what the agent should accomplish.
  </Step>

  <Step title="Save and open configuration">
    Select **Create Parent Agent**. You can now add sub-agents, tools, and triggers.
  </Step>
</Steps>

<Frame>
  <img src="https://mintcdn.com/maadify/X4aEKWAIWIr8YAhZ/images/parent-agents/create-parent-agent.png?fit=max&auto=format&n=X4aEKWAIWIr8YAhZ&q=85&s=feb1936576cb115ed7934e8969c5f124" alt="Create Parent Agent form showing Parent Agent Type, Action, Parent Agent Name, Parent Agent Name Key, and Description fields" width="1536" height="825" data-path="images/parent-agents/create-parent-agent.png" />
</Frame>

<Tip>
  Think of **Parent Agent Type** as the category and **Action** as the sub-category. Together they classify the agent so it's easier to find and reuse later.
</Tip>

## Parent agent configurations

Use **Parent agent configurations** to control default behavior for the entire flow. You can open the form from **Actions → Edit Parent Configuration** in the flow view.

<Steps>
  <Step title="Open the configuration form">
    In flow view, select **Actions → Edit Parent Configuration** (or **View Parent Configuration** if you do not have edit permission).
  </Step>

  <Step title="Update defaults">
    Change the behavior you want to apply across the flow, then save.
  </Step>
</Steps>

<Frame>
  <img src="https://mintcdn.com/maadify/X4aEKWAIWIr8YAhZ/images/parent-agents/parent_agent_configurations.png?fit=max&auto=format&n=X4aEKWAIWIr8YAhZ&q=85&s=916e331c0f489bd414ccca919cf6b10c" alt="Parent agent configuration showing tabs for sub-agents, rules, and triggers" width="726" height="980" data-path="images/parent-agents/parent_agent_configurations.png" />
</Frame>

### What each configuration does

* **Active**: turns the parent agent on or off.
* **Max conversation rounds**: caps how many agent-to-agent turns can occur in a single run.
* **Default agent on continued conversation**: chooses the first sub-agent when a conversation continues; if empty, standard routing rules are used.
* **Continue with last agent**: keeps the last agent for a continued conversation when no default agent is set.
* **Default LLM model**: fallback model when a sub-agent does not specify one.
* **Default temperature**: fallback creativity level when a sub-agent does not specify one.
* **Memory expiration settings**:
  * **Thread TTL (hours)**: short-term memory for the current thread.
  * **Agent TTL (days)**: medium-term memory across conversations for this agent.
  * **Tenant TTL (days)**: long-term memory across the organization.
  * **Clear thread on end**: clears thread memory when the conversation ends.

<Tip>
  Use defaults for consistency across sub-agents, then override at the sub-agent level only when you need different behavior.
</Tip>

## Go deeper

<CardGroup cols={2}>
  <Card title="Flow builder" icon="route" href="/agents/parent-agent-flow">
    Learn how nodes, edges, and rule order map to execution.
  </Card>

  <Card title="Sub-agents in flows" icon="bot" href="/agents/parent-agent-sub-agents">
    Add or edit sub-agents directly from the flow configuration.
  </Card>

  <Card title="Tool executions" icon="play" href="/agents/parent-agent-tool-executions">
    Configure payloads, loops, and templated tool inputs.
  </Card>

  <Card title="Maadify Browser Agent" icon="globe" href="/agents/browser-agent">
    Configure browser tools that agents can use to navigate, act, and extract data.
  </Card>

  <Card title="Rules and routing" icon="git-branch" href="/agents/parent-agent-rules">
    Define conditions, actions, and approvals that route workflows.
  </Card>

  <Card title="Triggers" icon="alarm-clock" href="/agents/parent-agent-triggers">
    Start workflows using schedules or webhooks.
  </Card>

  <Card title="Sub-agents" icon="bot" href="/agents/sub-agents">
    Build reusable sub-agents that plug into parent agents.
  </Card>

  <Card title="Prompt templates" icon="file-text" href="/agents/prompt-templates">
    Reuse prompts and variables across agents.
  </Card>
</CardGroup>
