# Chronicle Labs

> Build simulated systems, give your agent tasks, and score what it actually changed.

- [Chronicle](https://docs.chronicle-labs.com/introduction.md): Grade agents on what they do. Compare versions on repeatable tasks.
- [Quickstart](https://docs.chronicle-labs.com/platform/quickstart.md): Launch the example world, run your agent against its tasks, and follow the scores to the evidence.
- [Explore the example world](https://docs.chronicle-labs.com/product-explorer.md): Explore the Worldsmith workspace, follow recorded activity, and inspect tasks and scorers.
- [Build a world](https://docs.chronicle-labs.com/worldsmith/quickstart.md): Review an example world, launch its service twins, and prepare a task suite for your agent.
- [Write a good brief](https://docs.chronicle-labs.com/worldsmith/prompting.md): Give Worldsmith the starting evidence, the agent's job, and observable checks for success.
- [Service twins](https://docs.chronicle-labs.com/platform/twins.md): Simulated services that your agent can read and write through the vendor's API.
- [Connect your agent](https://docs.chronicle-labs.com/platform/agents.md): Connect an agent, run an evaluation, and compare results.
- [Tasks and scorers](https://docs.chronicle-labs.com/platform/tasks-and-scorers.md): A task is a job for the agent. A scorer checks the result. A task suite holds the tests you run again and again.
- [Run an evaluation](https://docs.chronicle-labs.com/platform/evaluations.md): Run agent versions against a task suite, compare them, and trace every score back to what the agent did.
- [Timeline](https://docs.chronicle-labs.com/platform/observability.md): See every request, change, and score. Turn what you find into the next test.
- [Concepts](https://docs.chronicle-labs.com/worldsmith/concepts.md): The basics of worlds, tasks, agents, and evaluation.
- [Environments](https://docs.chronicle-labs.com/platform/environments.md): The saved starting state every trial begins from. Worldsmith creates one for you; this page is for when you need more control.
- [Troubleshooting](https://docs.chronicle-labs.com/worldsmith/troubleshooting.md): Find the message you see, check its cause, and take the next action in Worldsmith.
- [Overview](https://docs.chronicle-labs.com/api-reference/overview.md): Build worlds, evaluate agents, and read their scores from code.
- [Authentication](https://docs.chronicle-labs.com/api-reference/authentication.md): One API key for everything in Chronicle. Twins have their own tokens.
- [SDKs](https://docs.chronicle-labs.com/api-reference/languages/overview.md): Official clients for Python, TypeScript, JavaScript, Go, and Rust.
- [Worlds](https://docs.chronicle-labs.com/api-reference/worldsmith.md): Compile a blueprint, launch a world, and get the twins and tasks it created.
- [Twins](https://docs.chronicle-labs.com/api-reference/twins.md): Find a twin's URL and token, call it like the real service, and read back what happened.
- [Agents](https://docs.chronicle-labs.com/api-reference/agents.md): Register agent versions, record what they do, and read the registry.
- [Tasks and scorers](https://docs.chronicle-labs.com/api-reference/tasks.md): Create a task suite, add tasks and scorers, and publish a frozen version to evaluate.
- [Evaluations](https://docs.chronicle-labs.com/api-reference/evaluations.md): Run agents against a task suite, wait for it to finish, and read the scores.
- [Events](https://docs.chronicle-labs.com/api-reference/telemetry.md): Send events, traces, and feedback into Chronicle, then query the history.
- [Errors and retries](https://docs.chronicle-labs.com/api-reference/reliability.md): Timeouts, retries, idempotency, pagination, and what an error tells you.
- [Python](https://docs.chronicle-labs.com/api-reference/languages/python.md): Install the Chronicle Python SDK and run your first evaluation.
- [TypeScript](https://docs.chronicle-labs.com/api-reference/languages/typescript.md): Install the Chronicle SDK and run your first evaluation from TypeScript.
- [JavaScript](https://docs.chronicle-labs.com/api-reference/languages/javascript.md): Install the Chronicle SDK and run your first evaluation from Node.js.
- [Go](https://docs.chronicle-labs.com/api-reference/languages/go.md): Install the Chronicle Go SDK and run your first evaluation.
- [Rust](https://docs.chronicle-labs.com/api-reference/languages/rust.md): Install the Chronicle Rust SDK and run your first evaluation.
- [List example worlds](https://docs.chronicle-labs.com/api-reference/worldsmith/list-example-worlds.md): Built-in example blueprints you can compile and launch as-is. Nothing is created by loading one.
- [Compile a blueprint](https://docs.chronicle-labs.com/api-reference/worldsmith/compile-a-blueprint.md): Validate a blueprint and preview the seed each twin would get, without launching anything. An invalid blueprint returns 200 with a non-empty problems list and no twins; fix the problems before launching.
- [Launch a world version](https://docs.chronicle-labs.com/api-reference/worldsmith/launch-a-world-version.md): Compile the blueprint, save a world version, and start one twin per service in the blueprint. Pass sourceWorldId to create a new version of an existing world. Reuse the same idempotencyKey to retry safely. Check status when the response arrives: failed comes with launchError. Worlds with tasks also…
- [List saved worlds](https://docs.chronicle-labs.com/api-reference/worldsmith/list-saved-worlds.md): Every saved world version in your organization, with its blueprint, twins, and status.
- [Get a saved world](https://docs.chronicle-labs.com/api-reference/worldsmith/get-a-saved-world.md): One saved world version with its status, twins (URL, token, status), and task setup in evaluation.
- [List world versions](https://docs.chronicle-labs.com/api-reference/worldsmith/list-world-versions.md): All versions of the same world, newest first. Each has its own world.id and twins.
- [Set up world evaluation](https://docs.chronicle-labs.com/api-reference/worldsmith/set-up-world-evaluation.md): Create or retry the world task suite, scorers, and environment version. No request body. Returns the world with the updated evaluation object. A world with no tasks returns 400.
- [List world activity](https://docs.chronicle-labs.com/api-reference/worldsmith/list-world-activity.md): Requests made to any twin in this world, newest first, each with the service, twin ID, and a summary of the records it changed. Use the twin activity detail endpoint for full payloads.
- [List world lifecycle events](https://docs.chronicle-labs.com/api-reference/worldsmith/list-world-lifecycle-events.md): World and twin lifecycle events, newest first: created, twin linked, seed stored, status changed, reset, stopped.

## OpenAPI Specs

- [worldsmith-openapi](/api-reference/worldsmith-openapi.json)