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

# Sub-agents

> Delegate scoped work to specialized, isolated agents

## Overview

Sub-agents are specialized agents you can delegate to. Each runs in its own isolated context with a restricted toolset, so the main conversation stays focused while a sub-agent handles a self-contained task — exploring the codebase, reviewing a change, or any role you define.

<Info>
  Delegate by @mentioning a sub-agent in the composer, or by using one as a node inside a [Workflow](/features/workflows). A sub-agent's full transcript opens in its own detail window.
</Info>

## Built-in Sub-agents

Parsaa ships with sub-agents out of the box:

| Agent           | How it's used                | Role                                                                                                        |
| --------------- | ---------------------------- | ----------------------------------------------------------------------------------------------------------- |
| `@tester`       | You @mention it              | Generates unit tests for the focused file — see [Test Generation](/features/test-generation)                |
| `@documenter`   | You @mention it              | Writes doc comments for the focused file — see [Documentation Generator](/features/documentation-generator) |
| General Purpose | Dispatched by the main agent | Handles any self-contained task when no specialized agent fits                                              |
| Explore         | Dispatched by the main agent | Fast, read-only investigation of the codebase                                                               |

<Note>
  **General Purpose** and **Explore** are dispatched automatically by the main agent — they don't appear in the @-mention picker. The workflow role agents (Developer, Reviewer, QA, Planner) are workflow-only — see [Workflows](/features/workflows).
</Note>

## What You Can Configure

Create and edit sub-agents in **Settings → Extensions → Agents**.

| Field          | Description                                                    |
| -------------- | -------------------------------------------------------------- |
| Name           | Canonical kebab-case identifier (e.g. `codebase-analyzer`)     |
| Display name   | Human-readable label shown in the UI                           |
| Role           | One-line description shown to the main model                   |
| System prompt  | The instructions that define the sub-agent's behavior          |
| Allowed tools  | The tool allowlist for this sub-agent (via a preset or custom) |
| Permissions    | Per-capability allow / ask / deny overrides                    |
| Model          | Optional model override (otherwise inherits the parent's)      |
| Max iterations | Cap on the sub-agent's steps per delegation                    |
| Icon & color   | Optional SF Symbol and accent color                            |

### Tool presets

Instead of hand-picking from \~26 tools, pick a preset for the agent's toolset:

| Preset           | Grants                                         |
| ---------------- | ---------------------------------------------- |
| **Read-only**    | Read & search — explore and review, no changes |
| **Read & write** | Read plus file edits                           |
| **Full access**  | Read, edit, build/run, git, and shell          |

A selection that doesn't match a preset shows as **Custom**. The full per-tool grid stays available under **Advanced → Tools** for fine-tuning. See [Tools & Permissions](/configuration/tools-and-permissions).

<Note>
  Sub-agents cannot dispatch other sub-agents — delegation is one level deep, which keeps runs predictable.
</Note>

## How Delegation Works

<Steps>
  <Step title="Create the sub-agent">
    Define its role, system prompt, allowed tools, and permissions in Settings → Extensions → Agents.
  </Step>

  <Step title="@mention it">
    In the composer, @mention the sub-agent to hand it a scoped task.
  </Step>

  <Step title="It runs in isolation">
    The sub-agent works with its own tools and context, then returns a result to the main conversation.
  </Step>

  <Step title="Inspect the transcript">
    Open the sub-agent detail window to see exactly what it did, step by step.
  </Step>
</Steps>

<Tip>
  Keep each sub-agent narrow — a tight role and a minimal toolset produce more reliable results than one do-everything agent.
</Tip>

## Related

<Card title="Workflows" icon="diagram-project" href="/features/workflows">
  Compose sub-agents into multi-step graphs with evaluators, breakpoints, and checks.
</Card>