> ## 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.

# Create risk categories

> Define the risk dimensions that risk profiles weigh.

A risk category is a named risk dimension, such as factual accuracy, built by pairing one or more [evaluators](/guard/app/evaluators) with the datasets they run against. Its aggregation method combines those scores into one category score, which a [risk profile](/guard/app/risk-profiles) then weights.

## Create a risk category

<Steps>
  <Step title="Create New Category">
    On the **Risk Categories** page, click **Create New Category**.
  </Step>

  <Step title="Create New Risk Category">
    * **Name** – required.
    * **Description** – required. What the category measures.
    * **Aggregation Method** – how the category combines its evaluators' scores: weighted average, minimum, maximum, median, or harmonic mean. The default is weighted average.
    * **Critical Threshold (optional)** – a score below which the category triggers high risk.
    * **Tags (comma-separated)** – optional labels for organizing categories.

    Click **Create Category**.
  </Step>
</Steps>

## Manage your risk categories

The **Risk Categories** table shows one row per category, with **Category Name**, **Description**, **Evaluators**, **Aggregation**, **Datasets**, **Used In Profiles**, **Visibility**, and **Actions**. Categories SeekrGuard provides carry a **System** badge, and each category's tags appear under its name.

Above the table, search by name, narrow the list with **Filter**, and filter by **Visibility**.

Each row's **Actions** menu offers **Edit**.

## Pair evaluators with datasets

Editing a category opens its **Evaluation Canvas**, where you decide what the category actually measures. Drag a dataset from **Datasets** onto the canvas, then drag evaluators from **Evaluators** onto that dataset. Each pairing scores those evaluators against that dataset, and the category's aggregation method combines the results into one score.

**Table View** shows the same pairings as a grid.

You can also change the category's **Description**, **Aggregation Method**, and **Visibility** from this page. Click **Save Changes** to keep them.