> For the complete documentation index, see [llms.txt](https://docs.umbraco.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.umbraco.com/ai-in-umbraco/testing-and-evaluation/tests.md).

# Overview

AI Testing and Evaluation framework for validating prompt and agent outputs.

The AI Testing and Evaluation framework (`Umbraco.AI` core) provides automated validation of AI outputs. You can define tests that run prompts or agents, then grade the results against success criteria to detect regressions and compare model configurations.

## Installation

The testing framework is included in the core `Umbraco.AI` package. No additional installation is required.

{% hint style="info" %}
You need the Prompt Management or Agent Runtime add-on installed to test prompts or agents respectively.
{% endhint %}

![The tests list showing configured tests with their features and run counts](https://583970477-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FCHKT9bQhbR3swAdBvil4%2Fuploads%2Fgit-blob-bf2052d50e2b90eb220f66b594f57d8abba8ece6%2Fbackoffice-ai-test-list.png?alt=media)

## Features

* **Graders** - Define success criteria with code-based and model-based graders
* **Variations** - A/B test across different models, profiles, and configurations
* **Baselines** - Set a run as a baseline for regression detection
* **Batch Execution** - Run multiple tests at once or filter by tags
* **Metrics** - Calculate pass\@k and pass^k for non-deterministic outputs
* **Transcripts** - Capture full execution traces for debugging

## How It Works

1. **Create a test** targeting a prompt or agent with one or more graders
2. **Run the test** to execute the target and grade the output
3. **Review results** including per-grader scores, metrics, and transcripts
4. **Set a baseline** and compare future runs for regression detection
5. **Add variations** to compare different model configurations side by side

## Quick Start

1. Navigate to the **AI** section > **Tests**
2. Click **Create** and select a prompt or agent to test
3. Add graders to define success criteria
4. Click **Run** to execute the test and review results

See [Getting Started](/ai-in-umbraco/testing-and-evaluation/getting-started.md) for a detailed walkthrough.

## Documentation

| Section                                                                     | Description                             |
| --------------------------------------------------------------------------- | --------------------------------------- |
| [Concepts](/ai-in-umbraco/testing-and-evaluation/concepts.md)               | Tests, graders, variations, and metrics |
| [Getting Started](/ai-in-umbraco/testing-and-evaluation/getting-started.md) | Step-by-step setup guide                |
| [Graders](/ai-in-umbraco/testing-and-evaluation/graders.md)                 | All built-in grader types               |
| [Variations](/ai-in-umbraco/testing-and-evaluation/variations.md)           | A/B testing across configurations       |
| [API Reference](/ai-in-umbraco/testing-and-evaluation/api.md)               | Management API endpoints                |

## Related

* [Prompt Management](/ai-in-umbraco/add-ons/prompt.md) - Prompt templates
* [Agent Runtime](/ai-in-umbraco/add-ons/agent.md) - Agent definitions
* [Guardrails](/ai-in-umbraco/concepts/guardrails.md) - Safety and compliance rules


---

# 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 by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.umbraco.com/ai-in-umbraco/testing-and-evaluation/tests.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.