> For the complete documentation index, see [llms.txt](https://help.usejimo.com/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.usejimo.com/docs/experiences/agent/knowledge/user-context.md).

# User context

Share selected user properties with the Agent so its answers fit the person asking, without exposing anything you did not choose to share.

**Dashboard path:** [Agent > Train > Knowledge > User Context tab](https://i.usejimo.com/agent/knowledge/user-context) (`/agent/knowledge/user-context`)

***

## What is User Context?

User Context is an allowlist of user properties the Agent is allowed to read when it answers. By default the Agent knows nothing about the user in front of it. Add a property here and the Agent receives its value at answer time, so it can adapt guidance to the user's plan, role, language or any attribute you track in Jimo.

<figure><img src="https://2794860263-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKzAcDWQbK1gKbpra7bkb%2Fuploads%2FobLAbHjPny5paQ1dCqKt%2Fimage.png?alt=media&amp;token=f6c95b22-f2d4-4ae5-b63f-527cc77b5917" alt=""><figcaption></figcaption></figure>

The property values are read from the same attributes you use for segmentation. Nothing new is collected: you choose which existing attributes the Agent may see. See [Attributes](/docs/analyze/users-and-segments/data-management/attributes.md) for where those values come from.

***

## Overview

* **Allowlist, not a firehose:** only the properties in the list reach the Agent. Everything else stays private.
* **Any attribute source:** Jimo SDK attributes, native attributes, or attributes from an integration.
* **Editable descriptions:** the display name and description you give a property are what the Agent reads to understand it, so a clear description improves the answers.
* **Applies to every answer in the live widget:** once a property is listed, the Agent takes it into account in all conversations in your product, including Skills and escalation hand-offs. The dashboard test chat does not read User Context.
* **Up to 20 properties, 500 characters each.** Longer values are cut.

***

## Use Cases

* 💳 **Plan-aware answers.** Share the `plan` attribute so the Agent stops suggesting features the user does not have.
* 🧑‍💼 **Role-aware guidance.** Share `role` so an admin and a viewer get different instructions for the same question.
* 🌍 **Locale and region.** Share `country` or `language` so the Agent adapts examples, currencies or wording.
* 🙋 **Personalised escalation.** The user's name and email travel with the hand-off, so your support team sees who is asking.

***

## How to Access

1. **Open the tab:** go to **Agent > Train > Knowledge** and click the **User Context** tab. \ <br>

   <figure><img src="https://2794860263-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKzAcDWQbK1gKbpra7bkb%2Fuploads%2FYSODbKnACxxZuajx4ca8%2Fimage.png?alt=media&amp;token=c406ba74-d4b4-4d15-9416-41c90440af78" alt=""><figcaption></figcaption></figure>
2. **Add a property:** click **+ Add property** and pick an attribute from your project. Use the **All types** filter to narrow by data type. \ <br>

   <figure><img src="https://2794860263-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKzAcDWQbK1gKbpra7bkb%2Fuploads%2F5KWtfXetpmAernDRFb8k%2Fimage.png?alt=media&amp;token=9c3b11b2-8673-4662-a053-9b5d473836f6" alt="" width="339"><figcaption></figcaption></figure>
3. **Describe it for the Agent:** open the row menu (**...**) and click **Edit attribute** to adjust the display name and the description the Agent will read. \ <br>

   <div><figure><img src="https://2794860263-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKzAcDWQbK1gKbpra7bkb%2Fuploads%2F5RbQkcLHbHvtXb2iTNud%2Fimage.png?alt=media&amp;token=26b399e7-87af-4e6e-a5fe-7fbbf25ae947" alt=""><figcaption></figcaption></figure> <figure><img src="https://2794860263-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKzAcDWQbK1gKbpra7bkb%2Fuploads%2FJYMI4mYU3gdfK0pADB87%2Fimage.png?alt=media&amp;token=3a36e58b-dbb6-4a7f-9318-cdf21b2015a3" alt=""><figcaption></figcaption></figure></div>
4. **Remove it when needed:** row menu, then **Remove from Agent Context**. The attribute itself is not deleted, only its access for the Agent. \ <br>

***

## Main Features

### The property list

<figure><img src="https://2794860263-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FKzAcDWQbK1gKbpra7bkb%2Fuploads%2FZ8ihgVFFIfxR9VwtPByE%2Fimage.png?alt=media&amp;token=b90ce07c-dfd7-41cb-a6fb-a796475b667d" alt=""><figcaption></figcaption></figure>

| Column           | What it shows                                                     |
| ---------------- | ----------------------------------------------------------------- |
| **Display Name** | The property's name and its description, as the Agent reads them. |
| **Data Type**    | Text, Number, Boolean, Date or List.                              |
| **Source**       | Where the value comes from: Jimo SDK, Native, or an integration.  |

The **Search property** field filters the list by name.

### Test with real context

User Context is read in the live widget only: the dashboard test chat does not receive it.

To check that the Agent adapts its answer once a property is added, open the Agent in your product as a user who has that property, in a new chat.

{% hint style="warning" %}
Only share what the Agent needs to answer. A property in this list is sent to the language model with every conversation. Keep personal data out unless it changes the answer.
{% endhint %}

***

## Best Practices

* **Start with two or three properties.** Plan, role and language cover most cases.
* **Write the description as an instruction.** "The user's subscription plan: free, pro or enterprise. Do not recommend enterprise features to free users." works better than "Plan".
* **Pair with General Instructions.** Tell the Agent in [Chat](/docs/experiences/agent/chat.md) how to use the properties it receives.
* **Review with your privacy owner.** The list is short and visible, which makes it easy to audit.

***

## Related pages

* [Knowledge](/docs/experiences/agent/knowledge.md): the four tabs of Agent knowledge.
* [Attributes](/docs/analyze/users-and-segments/data-management/attributes.md): manage the underlying attributes.
* [Escalation](/docs/experiences/agent/escalation.md): what your support team receives with a hand-off.