> ## 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.
# Sub-Agents
> Manage reusable sub-agents from the Sub Agents tab outside of any parent agent flow
## What the sub-agent library is
The **Sub Agents** tab on the **Agent Configurations** page lists every sub-agent your tenant can use. Sub-agents are reusable across parent agents, so editing one updates every flow that includes it. Use this view when you want to create a sub-agent up front, edit one that is shared across flows, or audit which parent agents and tools are attached to it.
For adding or editing a sub-agent inside a single parent agent flow, see [Sub-agents in Parent Agent Flows](/agents/parent-agent-sub-agents). Both entry points open the same dialog and expose the same fields.
## Open the Sub Agents tab
Go to **Agent Configurations** from the sidebar.
At the top of the page, select **Sub Agents** in the segmented control next to the heading. The page heading changes to **Sub Agent Configurations** and the grid of sub-agents loads below.
The **Sub Agents** tab only appears if your role includes the **sub\_agents.read** permission. The **Add Sub Agent** button is gated on **sub\_agents.create**, and inline edits require **sub\_agents.update**.
## Browse the sub-agent grid
The grid shows one card per sub-agent and supports search, filtering, and pagination.
* **Search Sub Agents**: type a name or keyword and press **Enter** or click **Search** to filter the list.
* **Show Shared Sub Agents**: toggle on to include sub-agents owned by other tenants that have been shared with you. Toggle off to see only sub-agents your tenant owns.
* **Pagination**: when there are more results than fit on one page, use the pager controls below the grid to move between pages.
### What each card shows
Each card summarizes a sub-agent at a glance so you can decide whether to open it.
* **Agent type badge**: **Assistant** (green), **Proxy** (purple), **User Proxy** (purple), or **Custom** (blue).
* **Owner badge**: the tenant that owns the sub-agent, or **System** for built-in sub-agents.
* **Name and icon**: the sub-agent display name with a bot icon for assistants and a code icon for proxies.
* **Description**: a short summary of what the sub-agent is for. Hover to see the full description.
* **System Prompt**: the prompt template attached to the sub-agent. Use **Add**, **Edit**, or **View** (depending on your permissions) to manage the template inline without leaving the grid.
* **Tools**: overlapping connector avatars for each tool the sub-agent can call. Hover an avatar to see the tool and connector name. A **+N** badge appears when there are more tools than fit; hover it to list the rest. Click any tool avatar to jump straight into editing that tool on the sub-agent.
* **Parent Agents**: avatars for every parent agent that uses this sub-agent. Hover for the parent agent name, action, tenant, and description, and click an avatar to open that parent agent in a new tab.
## Create, edit, or copy a sub-agent
From the grid you can:
* **Add a new sub-agent** by selecting **Add Sub Agent** in the top-right of the grid. The dialog opens in create mode.
* **Edit an existing sub-agent** by clicking its card. If your tenant owns the sub-agent and you have **sub\_agents.update**, the dialog opens in edit mode; otherwise it opens in view-only mode.
* **Create a copy of a shared sub-agent** by opening it and toggling **Create a Copy of this Agent** before saving. Copies are owned by your tenant and can be edited without affecting the original.
Copy a shared sub-agent when you need a tenant-specific change to its prompt, tools, or memory configuration. Edits made on the copy do not affect the parent agents that still use the original.
For details on every field in the dialog (identity, LLM, visibility, memory, structured output), see [Sub-agents in Parent Agent Flows](/agents/parent-agent-sub-agents#sub-agent-fields-you-can-edit).
## Structure sub-agent outputs
Assistant sub-agents can return structured JSON instead of free-form text. Use structured output when you want the sub-agent to extract values, classify a request, summarize records into fields, or pass predictable data into a later workflow step.
The schema builder helps you define the shape of the response. You can:
* **Add fields manually** with names, display names, descriptions, and types.
* **Choose field types** such as string, integer, number, boolean, object, arrays, and object arrays.
* **Mark fields as required** so the agent must include them in the response.
* **Add options** when the agent should choose from a fixed list.
* **Set defaults, ranges, and length limits** for supported values.
* **Create nested properties** for objects and object arrays.
* **Generate fields** from a JSON schema or from an example JSON response.
Structured outputs are useful because later workflow steps can reference fields directly. For example, a sub-agent can return `priority`, `customer_name`, `line_items`, or `approval_summary`, and a parent agent can use those values in rule conditions, tool inputs, approvals, and notifications.
Changing a structured output schema can affect parent agents that map those fields into rules, tool inputs, approvals, or notifications.
## Edit impact
Editing a sub-agent updates every parent agent that uses it. Review dependent agents before saving changes.
Changing prompts or tools can alter live workflows. Validate with a test run.