---
type: Document
title: Agents Menu
description: Short presentation of the Agents menu for running agents configured in the Workspace.
tags:
  - agents
  - workspace
  - execution
  - llm
  - run
  - automation
  - documentation
timestamp: 2026-07-01T17:04:00Z
status: Draft
---

### **Agents** Menu

The <kbd>Agents</kbd> menu lets you quickly run the agents you configured in the <kbd>Workspace</kbd>.

The principle is simple:

- <kbd>Workspace</kbd> is used to create and configure agents
- <kbd>Agents</kbd> is used to run them

The <kbd>Agents</kbd> menu is available from any page in **Living Documentation**: Home, Workspace, Admin, Files, Diagrams, and so on.

<!-- image-width: 1/3 -->

![Agents menu listing available agents by LLM provider](/images/DOCUMENTATION/execution_d_agents.png)

---

### What you see in the menu

Agents are grouped by LLM provider.

<!-- layout-columns -->

<!-- col -->

Each group corresponds to a model or provider configured in the Workspace.

When you open a group, you see the agents available for that provider.

<!-- col -->

| Element                  | Meaning                                       |
| ------------------------ | --------------------------------------------- |
| LLM provider             | Model used to run the agent                   |
| Agent                    | Available documentation task                  |
| Required input           | The agent expects information before starting |
| Incomplete configuration | The agent cannot be launched yet              |

<!-- /layout-columns -->

<!-- quote-type: info -->
<!-- quote-title: Info -->
<!-- quote-icon -->

> If an agent does not appear in this menu, check that it has been created in the <kbd>Workspace</kbd> and attached to an LLM provider.

---

### Run an agent

To run an agent:

1. click <kbd>Agents</kbd>
2. open the desired LLM provider
3. click the agent
4. fill in the input if the agent asks for one
5. start the execution

Some agents start immediately.

Others ask for input, for example:

- a document id
- a question
- an image prompt
- text to transform
- a one-off instruction

---

### Example with input

<!-- layout-columns: 1/2 | gap: sm -->

<!-- col -->

When an agent asks for information, **Living Documentation** displays a popup.

The input depends on the agent.

In this example, the agent expects a full prompt to generate an image.

<!-- col -->

<!-- image-width: 2/3 -->

![Popup asking for the input required before running an agent](/images/DOCUMENTATION/popup_execution_agent.png)

<!-- /layout-columns -->

---

### During execution

While the agent works, **Living Documentation** displays progress toasts, so you can keep browsing **Living Documentation** while the agent is running.

<!-- layout-columns -->

<!-- col -->

Depending on the configuration, you may see:

- the model call
- the tools being used
- intermediate results
- a retry without tools if the model does not accept them

You do not need to stay blocked on the popup: the execution produces a result document.

<!-- col -->

<!-- image-width: full -->

![Toast showing real-time progress for an agent execution](/images/DOCUMENTATION/feedback_execution_agent.png)

<!-- /layout-columns -->

---

### The result document

Each execution creates a run document that you can access through a direct link from the execution toast.

<!-- image-width: 2/3 -->

![Success toast containing the link to the document generated by the agent](/images/DOCUMENTATION/summary_agent_execution.png)

<!-- layout-columns: 1/3 -->

<!-- col -->

This document generally contains:

- the agent that was used
- the provider that was used
- the model that was used
- the execution status
- the user input
- the agent response

<!-- col -->

![Run document created after an agent execution](/images/DOCUMENTATION/run_agent_execution.jpg)

<!-- /layout-columns -->

If the <kbd>Debug agents</kbd> option is enabled in <kbd>Admin</kbd>, the document also contains the technical details of the execution, especially the conversations between the tools and the agent during agentic iterations.

---

### When should you use this menu?

Use <kbd>Agents</kbd> when you want to apply a task that is already prepared.

Examples:

- translate a document
- correct a document
- summarize a note
- generate an image
- transform text into a table
- improve existing Markdown
- produce a documentation draft

<!-- quote-type: success -->
<!-- quote-title: Quick Win -->
<!-- quote-icon -->

> A good agent does one thing, but does it often and cleanly.

---

### Simple advice

Start with two or three very useful agents.

For example:

- a Markdown correction agent
- a translation agent
- a documentation image generation agent

When they are reliable, gradually add more specialized agents.
