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

# Centurion

> Centurion is the master orchestrator of the Myrmex AI agents — it interprets your request, routes it to the right specialists, and coordinates their work. Its SOC mode drives alerts and investigations.

import { CardGroup, Card, Note, Tip } from '@mintlify/components';

**Centurion** is the master orchestrator of the Myrmex AI agents and your single point of contact in the [Workspace](/documentation/ai-workspace/overview). When you send a request without naming anyone, Centurion is who reads it, decides what needs to happen, and puts the right specialists to work.

## What Centurion Does

<CardGroup cols={2}>
  <Card title="Understands your intent" icon="brain">
    Reads your request in plain language and works out what you actually want to
    achieve — whether you're a hands-on technician or a results-focused manager.
  </Card>

  <Card title="Plans the approach" icon="map">
    For anything beyond a single step, Centurion decides which specialists are
    needed and in what order, instead of acting blindly.
  </Card>

  <Card title="Routes to specialists" icon="users">
    Dispatches work to the right agents — Perseus on a host, Hydra on an
    integration, Orion for threat intel, Scribe for a report — and keeps them
    coordinated.
  </Card>

  <Card title="Synthesizes results" icon="lightbulb">
    Brings the specialists' findings back together into one clear answer, so you
    don't have to stitch the pieces yourself.
  </Card>
</CardGroup>

## The Router at the Center

There is no model or agent selector in Myrmex — you direct the team with `@`-mentions. When you `@`-mention a specialist, your request goes straight to them. When you mention **no one**, Centurion decides who should handle it and routes accordingly. Either way, it stays aware of the full [context](/documentation/get-started/concepts) you're working in and the entities you've referenced.

Centurion coordinates the work, but it never bypasses your control. Actions that change your environment follow the **AI proposes, you approve** pattern — Centurion presents the plan, and you decide what runs.

## SOC Mode: Alerts and Investigations

Centurion has a dedicated **SOC variant** that powers detection and response work. When you investigate an [alert or case](/documentation/detection-response/overview), the platform composes a security-focused prompt — with the relevant indicators already extracted — and hands it to Centurion's SOC mode. From there it coordinates the same specialists to enrich indicators, inspect affected hosts, and recommend containment, then writes the results back to you.

<Note>
  Centurion is one of the AI agents — separate from the **software agent**
  installed on your hosts. Centurion *directs* that software agent through
  specialists like [Perseus](/documentation/multi-agent-system/agent-endpoint-specialist);
  it doesn't run on your endpoints itself.
</Note>

## Where to Go Next

<CardGroup cols={2}>
  <Card title="The AI agents" icon="users" href="/documentation/multi-agent-system/overview">
    Meet the full roster Centurion coordinates.
  </Card>

  <Card title="Detection & Response" icon="bell" href="/documentation/detection-response/overview">
    See Centurion's SOC mode at work on alerts and investigations.
  </Card>
</CardGroup>

<Tip>
  You rarely need to name Centurion — just ask. It's the default handler for any
  request where you don't `@`-mention a specific specialist.
</Tip>