> For the complete documentation index, see [llms.txt](https://docs.duplocloud.com/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.duplocloud.com/docs/armor/agents/agent-memories.md).

# Agent Memories

Agent Memories give your DuploCloud AI agents **persistent, shared knowledge about a workspace** — facts, conventions, and preferences that the agent should already know every time it works on a ticket in that workspace, without you having to repeat them.

Unlike a single ticket's conversation history, a memory is:

* **Shared** — every ticket created in the workspace can use it, not just the one where it was written.
* **Persistent** — it doesn't expire or get summarized away. It's a plain Markdown file that stays until you edit or delete it.
* **Curated by you** — memories are written and maintained by admins, not silently inferred by the agent.

Each memory is a small Markdown file with a `type` that tells the agent (and other admins) what kind of knowledge it holds:

| Type        | Use it for                                                  |
| ----------- | ----------------------------------------------------------- |
| `user`      | Who the user is — role, preferences, expertise              |
| `feedback`  | How to work — corrections, confirmed approaches             |
| `project`   | Ongoing work, decisions, deadlines                          |
| `reference` | Pointers to external resources — dashboards, tickets, links |

Every workspace also has one special file, `MEMORY.md`, which is an index — a one-line summary of every memory the agent loads on every ticket. The full content of an individual memory file is only pulled in when the agent decides it's relevant.

***

## Step 1 — Open the Memories admin page

Memories are managed from **AI Admin**, not from inside a specific agent's workspace. Click the app switcher in the top left (it shows your current app, e.g. **AI DevOps**) and select **AI Admin**.

![](/files/XRF0srySndbEP8s8wPdW)

![](/files/Hb3OzECvTKBbtVeeCOLy)

In **AI Admin**, select **Memories** from the left sidebar.

![](/files/xKp85ePH14nCh3JQ1Q5x)

## Step 2 — Pick a workspace

The Memories page shows one workspace's memory files at a time. Use the **Workspace** dropdown at the top to choose which workspace you want to manage.

![](/files/cOwIjJdsSR4dFReyfIcV)

![](/files/WQ0mvHFcMg4seODLP1st)

## Step 3 — Browse existing memories

The **FILES** panel on the left lists every memory file in the workspace. Click `MEMORY.md` to see the index the agent loads on every ticket — one line per memory, with its description and whether it was user-created or user-edited.

![](/files/su2x3U6Zr00HaEuXWE95)

Click any other file to view or edit its full content in the editor on the right.

![](/files/4sFbfhxndDKirmuR1qGB)

## Step 4 — Create a new memory

Click the **⋮** menu on the `memory` folder and choose **New memory**.

![](/files/oie7k5V1dXe8oMfJO6o4)

![](/files/3WOQMf8sLRnaKKpZL4pR)

Fill in the form:

* **Name** — lowercase, hyphen-separated; this becomes the filename (`<name>.md`).
* **Description** — the one-line summary shown in `MEMORY.md`.
* **Type** — hover the info icon for a reminder of what each of the four types means.
* **Memory** — the actual content the agent should know.

![](/files/CMgATIExNzU0HH0enoJe)

![](/files/BTtotarpTswL844QfJ11)

![](/files/tpLgMRSoGlZeH18BLvRZ)

![](/files/46HR0inYcxmY9Cvf8lGK)

Click **Create**. The new file is saved, added to the `memory` folder, and automatically appended to the `MEMORY.md` index.

![](/files/IjkM68tPYQLXBsn6U7kX)

![](/files/r06hybY0bYmtxnHGcJav)

{% hint style="info" %}
Editing or deleting a memory works the same way — open the file to edit it and click **Save**, or use its **⋮** menu to **Download** or **Delete** it. Deleting a memory also removes its line from `MEMORY.md`.
{% endhint %}

## Step 5 — Turning memory off for a single ticket

Workspace memory is used by default, but any individual ticket can opt out. When creating a ticket, expand **Advanced Options** and use the **Workspace Memory** toggle.

![](/files/AB42xbnqd8pA1oVp2qC8)

This is useful for one-off tickets where you deliberately don't want the agent to bring in the workspace's usual context.

## Step 6 — Seeing a memory in action

Here's the `deploy-conventions` memory created above, being used on a real ticket. With **Workspace Memory** left on, a new ticket asks the agent about the team's deployment process:

![](/files/0HLzKdCz8FWn6wvxpLAd)

The agent reads the memory and answers directly from it, without needing to be told the deployment process again:

![](/files/xFwstT6LOpKDuTbgkhZf)

The same **Workspace Memory** toggle is available on any open ticket's **Details** panel, so you can check or turn off memory for that specific conversation at any time:

![](/files/g6ZVLxChP86pnE1xDQp3)


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.duplocloud.com/docs/armor/agents/agent-memories.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.