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

# Concepts

> The basics of worlds, tasks, agents, and evaluation.

Worldsmith builds the simulated system where you test your agent.

## Four building blocks

<div className="chronicle-definition-list">
  <div><p><strong>World</strong></p><p>A simulated system with services, data, and starting state.</p></div>
  <div><p><strong>Task</strong></p><p>A job for your agent, with scorers that check success.</p></div>
  <div><p><strong>Agent</strong></p><p>The AI agent you want to test.</p></div>
  <div><p><strong>Evaluation</strong></p><p>A run that tests agents against tasks. Each attempt is a trial. In the dashboard, evaluations are called <strong>Backtests</strong>.</p></div>
</div>

## How agents are graded

Scorers check the agent's actions, changed records, or final answer using code
or an LLM judge. Only attached scorers affect the score, so add a check for each
requirement you care about.

Compare scores across agent versions to see what improved or regressed.
[Learn about tasks and scorers](/platform/tasks-and-scorers).

## Build, launch, evaluate

1. **Build:** describe your system and review the world, tasks, and scorers.
2. **Launch:** save a version and start its service twins—the simulated services your agent can call.
3. **Evaluate:** run your agent, then inspect its scores and Timeline.

Each trial starts fresh from the saved environment. Draft edits take effect
when you launch a new version.

<Card title="Build a world" icon="wand-magic-sparkles" href="/worldsmith/quickstart">
  Create a world and test your agent.
</Card>