> ## Documentation Index
> Fetch the complete documentation index at: https://docs.seekr.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Navigate the app

> Get oriented on the Home page and the navigation sidebar.

## Start on the Home page

When you open SeekrGuard, the **Home** page shows a **Welcome** panel of quick-start cards that link to **Examinations**, **Datasets**, **Evaluators**, and **Models**. Dismiss the panel with the **×** in its top-right corner. It stays hidden in later sessions.

Below the panel, the **Model Risk Assessment** section ranks your registered models by risk under a single profile.

* **Sort By** – order the models by risk score or by any of the profile's category scores, and use the button beside it to switch between ascending and descending.
* **Risk Level** – show only models at one risk level, or all.
* **Profile** – choose the risk profile that scores the models. The selector shows whether the profile is active and how many categories it contains.

Each model appears as a numbered card with its risk level badge, risk score, and some of the profile's category scores. **View all metrics** opens the profile page with the full set.

When a category score falls below its threshold, the card shows a count of threshold violations. Point to the count to see each category with its score and threshold.

**View Model Card** opens the model's card, **Test Model** starts a new examination, and **Chat with Model** opens the Chat page. Clicking the card itself selects the model and scopes the **AI Assistant** panel to it.

> [!NOTE]
>
> The risk score is a measure of risk, where higher is worse. The category percentages beneath it are performance scores, where higher is better. A model with high category percentages therefore carries a low risk score. See [how a risk score is calculated](https://docs.seekr.com/guard/app/risk-profiles.md#understand-how-a-risk-score-is-calculated).

The buttons at the top right of the section switch between three views:

* **Cards** – one card per model, ranked by the sort you choose.
* **Radar** – all models on a single chart, each drawn as a polygon across the profile's categories on a 0 to 100 scale.
* **Bars** – one bar per model for the metric you pick from the **Metric** list, either the overall risk score or one of the profile's categories. Bar color reflects the model's risk level.

To hide a model from the dashboard, use **Hide from dashboard** on its card or click its name below the radar chart. It stays hidden in later sessions. A count of hidden models and **Show all** then appear beside the view buttons.

## Ask the AI assistant

The floating  button opens the **AI Assistant** panel. It is available on the Home page and on an examination's results.

Clicking a model's card, or its bar in a chart, scopes the assistant to that model. The panel header then shows what it is scoped to, with a chip per model and **Clear** to reset the selection. The `{}` button clears the conversation.

## Use the navigation sidebar

The sidebar sections, in order, are:

| Section | Purpose |
| - | - |
| **Home** | Review model risk under a profile and open quick-start tasks. |
| **Chat** | Talk to a model and evaluate its responses inline. |
| **Risk Profiles** | Create and manage weighted sets of risk categories. |
| **Risk Categories** | Define the individual risk dimensions used by profiles. |
| **Examinations** | Create and review batch evaluations. |
| **Comparisons** | Score one response with several evaluators and judge models. |
| **Evaluators** | Browse the evaluator library and create custom evaluators. |
| **Datasets** | Upload and manage test datasets. |
| **AI-Ready Data** | Generate evaluation datasets from source documents. |
| **Models** | Register and manage models. |


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.