> ## 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.

# Agent Memory

> A short note the AI keeps current on each device and integration, so it remembers what matters about an asset from one interaction to the next.

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

Every device and every integration in Myrmex carries a small **Memory** — a running note about that specific asset that the AI keeps up to date on its own. It's how the agents remember the things that matter about a system between conversations, without you having to re-explain them each time.

<Note>
  Memory is maintained by **Centurion**. It's **read-only** in the console — you
  review it, and the AI writes and refreshes it as it learns while managing the
  asset. If you want to *instruct* the AI about an asset, use its **Rule** instead
  (see below).
</Note>

## What It Holds

Memory is a short, plain-language summary of the important points about managing one asset — the kind of context a teammate would jot down so the next person isn't starting from scratch. For example:

* *"This device is part of the Myrmex project."*
* *"Primary domain controller — changes here need extra care."*
* *"This firewall's management interface is only reachable through the DMZ collector."*

As the agents work with the asset, Centurion updates the note when it learns something worth remembering, keeping it concise rather than logging everything.

## Memory vs. Rule

Each asset has two adjacent, complementary notes — you'll find both on the asset's **Rules** tab:

<CardGroup cols={2}>
  <Card title="Rule — you instruct" icon="scroll">
    A free-text instruction **you** write for the AI — *"prefer PowerShell on this
    host,"* *"never restart this server during business hours."* The agents follow
    it whenever they act on the asset.
  </Card>

  <Card title="Memory — the AI remembers" icon="brain">
    A note **the AI** maintains about the asset as it learns. Read-only to you —
    context the agents carry forward, not an instruction you set.
  </Card>
</CardGroup>

## Where to Find It

Open a [device](/documentation/agent-endpoint-security/device-details) or an integration and go to its **Rules** tab — Memory sits just below the Rule. It travels with the asset, so whenever you (or the agents) work on that device or integration, its memory is part of the picture.

<Tip>
  Because memory is per-asset, `@`-mentioning a device or integration in chat brings
  the AI's understanding of *that* asset into the conversation. Pair it with a
  [folder rule](/documentation/ai-workspace/folders-and-rules) when you want
  guidance that applies to a whole group.
</Tip>

## Related

<CardGroup cols={2}>
  <Card title="Folders & rules" icon="folder-tree" href="/documentation/ai-workspace/folders-and-rules">
    Steer the AI across a whole group of assets with folder-level rules.
  </Card>

  <Card title="Device details" icon="server" href="/documentation/agent-endpoint-security/device-details">
    Where a single host's system info, rule, and memory live.
  </Card>
</CardGroup>