> For the complete documentation index, see [llms.txt](https://docs.interactive.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.interactive.ai/improvement/annotations.md).

# Annotations

Annotation is where people **score the agent's work by hand**. An evaluator can judge thousands of outputs in minutes, but it cannot tell you what good means in your business. A reviewer can, and what they leave behind becomes the reference the automated scores are measured against.

A **queue** is a batch of work for a reviewer. It carries a name, a description telling reviewers what to look for, and the [score configs](/settings/score-configs.md) that define which dimensions they will score. Those configs have to exist first, since a queue with no dimensions has nothing to collect.

<div data-with-frame="true"><figure><img src="/files/gMXD19onRGVDbADBYpjQ" alt=""><figcaption></figcaption></figure></div>

### Filling a queue

**Queues hold traces**, and you add them from Observability: one at a time from a trace's own Annotate control, or in bulk by selecting rows and using **Add to Annotation Queue**. Filter to what you want reviewed before selecting.

<div data-with-frame="true"><figure><img src="/files/QG88UqeyZQbylAUyGyh1" alt=""><figcaption></figcaption></figure></div>

Each item sits Pending until someone completes it, and the queue records who completed each one, so progress across a team is visible without having to ask.

### Reviewing

Process queue opens the reviewer's view: the item on one side, the scoring panel on the other, and Complete and Next to save and move on.

<div data-with-frame="true"><figure><img src="/files/U43366AtJX7wGDcjADvb" alt=""><figcaption></figcaption></figure></div>

There are two layouts. **Focused** shows the input, the output, and the metadata, which is enough for most judgments and keeps a reviewer moving. **Detailed** shows the whole trace alongside, with its observation tree, cost, latency, and Scores tab, for when the judgment depends on what actually happened rather than only on what came out.

The **scoring panel** takes its shape from the queue's score configs. A categorical dimension offers its labels, a numeric one takes a value, a boolean one offers true or false. A reviewer can only enter what the config allows, which is what keeps scores from ten different people comparable.

### Where the scores go

Annotation scores land on the **trace's Scores tab** and in Scores, tagged with annotation as their source. That is what lets you hold a reviewer's judgment next to an evaluator's on the same trace and see where the two disagree.

<div data-with-frame="true"><figure><img src="/files/bpfDixGucuEor8OoS0DP" alt=""><figcaption></figcaption></figure></div>


---

# 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 dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.interactive.ai/improvement/annotations.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

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.