> ## 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.

# Core Concepts

> The building blocks of the Myrmex platform: organization hierarchy, contexts, the Workspace, and how the agents work

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

Myrmex's orchestration and Artificial Intelligence (AI) capabilities, embodied in **Centurion**, represent the evolution of cybersecurity. This is not just about automating tasks — it's about infusing strategic intelligence and adaptability into every layer of your digital defense. Before diving in, it helps to understand a few core concepts that everything else builds on.

## How Your Account Is Organized

Myrmex uses a three-level hierarchy. Understanding it makes everything else — permissions, data isolation, and switching environments — click into place.

<CardGroup cols={3}>
  <Card title="Domain" icon="globe">
    Your company's top-level space in Myrmex, tied to your email domain. Created
    once, during first access.
  </Card>

  <Card title="Organization" icon="building">
    A tenant inside your domain. Enterprise and MSSP customers can create
    **sub-organizations** to manage multiple clients or business units, each
    isolated from the others.
  </Card>

  <Card title="Context" icon="layer-group">
    A working environment inside an organization — for example "Production",
    "Datacenter SP", or a specific client. Devices, integrations, alerts, and
    reports all belong to a context.
  </Card>
</CardGroup>

<Note>
  Almost everything you see in Myrmex is **scoped to the context you have
  selected**. Switching context filters the data, assets, and permissions to that
  environment. See [Multi-Context](#multi-context) below.
</Note>

## The Workspace

The **Workspace** is where you work with Myrmex: a conversational surface backed by the AI team, a navigable **Directory** of everything in your environment, and a tabbed area where reports, devices, integrations, playbooks, and dashboards open side by side. The concepts below are the resources you'll use inside it.

### Context

A device or integration you attach to a request gives the platform explicit, persistent context about a specific resource in your environment. It's the bridge between the agents' intelligence and the operational reality of your infrastructure — so a question or action targets exactly the right asset.

### Attachments

Attaching files lets you provide documents and data directly to the platform. This empowers the agents to analyze the content to answer your questions or enrich the task context, making the interaction richer and more precise.

### Rules

A rule on a device or integration lets you define specific instructions for how agents interact with that resource. It guides operational behavior (for example, "Use PowerShell for all commands"), optimizes execution, and ensures policy compliance — customizing how the platform operates in your environment.

### Memory

Memory lets the Myrmex platform remember past interactions, preferences, and long-term contextual information. This ensures personalized, continuous, and intelligent responses — avoiding repetition and building a relevant history for your needs.

### Directory

The Directory is the main tool for exploring and accessing every area of the Myrmex platform. It's designed for an intuitive, efficient experience, letting analysts move quickly between different operational contexts and assets.

### Multi-Context

The context and organization selector lets you switch between work environments, filtering data and permissions for the selected entity and ensuring information is kept segregated.

## Talking to the Agents

You direct work in the Workspace with natural language. To be precise, use `@`-mentions:

<CardGroup cols={2}>
  <Card title="Mention an agent" icon="at">
    Route a request to a specific specialist — for example ask Perseus to run a
    check on a host, or Scribe to write a report. If you don't mention anyone,
    Centurion picks the right agent for you.
  </Card>

  <Card title="Mention a resource" icon="tag">
    Scope a request to a specific device, integration, alert, report, folder, or
    file. Mentioned resources appear as chips so the agents know exactly what
    you're referring to.
  </Card>
</CardGroup>

## AI Proposes, You Approve

Myrmex is built around a human-in-the-loop model. When the AI generates something consequential — an automation playbook, a change request, or a risk acceptance — it's created as a **draft** or **proposal** that a human reviews and approves before it takes effect. You stay in control of what actually runs in your environment.

<Note>
  This pattern shows up across the platform: playbooks are generated as drafts,
  change requests (GMUD) require approvals, and risk acceptances are signed off
  before a finding is silenced.
</Note>