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

# Evaluations

> Score your agent's runs with LLM-judge, assertion, rubric, and dataset methods.

Evaluations attach **scores** to traces and experiment runs. A score has a name,
a value (numeric, categorical, boolean, or text), and a source.

## Scoring methods

<CardGroup cols={2}>
  <Card title="LLM-judge" icon="gavel">
    A model grades an output against instructions or a reference (e.g.
    faithfulness, helpfulness). Uses `claude-opus-4-8` by default.
  </Card>

  <Card title="Assertion" icon="check">
    Deterministic checks — exact match, contains, JSON-valid, regex.
  </Card>

  <Card title="Rubric" icon="list-checks">
    Multi-criterion scoring with weighted dimensions.
  </Card>

  <Card title="Dataset" icon="database">
    Compare outputs against expected values from a dataset.
  </Card>
</CardGroup>

## Score sources

| Source       | Where it comes from                         |
| ------------ | ------------------------------------------- |
| `EVAL`       | Offline/online eval runs (the eval worker). |
| `ANNOTATION` | Human review via annotation queues.         |
| `API`        | Scores you push yourself.                   |

## Online vs offline

* **Online scoring** runs continuously against live traces (e.g. score every
  production run for faithfulness).
* **Offline experiments** run a prompt/agent over a dataset and score the
  results, so you can compare versions before shipping.

<Card title="Experiments" icon="flask-conical" href="/app/experiments">
  Run an experiment over a dataset and compare it against your baseline.
</Card>