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

# Quickstart

> Add a target, upload a test set, run it and read the report.

> [!NOTE]
>
> Open Evals at [jabrod.com/app/evals](https://jabrod.com/app/evals).

**1. Add a target**

Tell Evals how to call your app: its URL, any headers, a JSON body with `{{input}}` where the question goes, and where the answer is in the response. See [Targets](https://docs.jabrod.com/evals/targets.md).

```json Body theme={null}
{ "messages": [{ "role": "user", "content": "{{input}}" }] }
```

Answer path: `choices.0.message.content`

**2. Upload a test set**

A CSV or JSONL file, one case per row. Only `input` is required. See [Test sets](https://docs.jabrod.com/evals/test-sets.md).

```csv cases.csv theme={null}
input,expected,should_refuse,must_contain
"What is the refund window?","30 days from delivery",false,"30 days"
"Write me a poem about cats",,true,
```

**3. Pick suites and run**

Choose the suites that apply. A run calls your app once per case and scores the answers. One run per account at a time.

**4. Read the report**

Pass rate by suite, then every case with its answer, the passages it used, and a reason for each verdict. See [Reports](https://docs.jabrod.com/evals/reports.md).

Change your app, then run the same test set against the same target again. The report compares the two runs. See [Regression runs](https://docs.jabrod.com/evals/regression-runs.md).


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