What it is
WebArena supplies reproducible websites and natural-language tasks for autonomous web agents. Tasks involve browsing and interacting with applications such as shopping, forums and content management.
It tests whether an agent can reach a useful outcome across a sequence of web interactions, including tasks that span more than one application.
How an agent works in it
Sees
Browser content, with observation modes defined by the harness.
Does
Navigate, click, type and interact with website controls.
Feedback
Task evaluators check the resulting application state or answer.
Example task (illustrative): Find an item that meets several constraints and update its listing.
Setup
Self-host the supplied application stack and use the evaluation harness. Website state must be reset consistently between tasks.
Use it with BoltzLabs
BoltzLabs does not host this project. To record an agent working on one of its tasks:
- Prepare one task from the project: its starter files, instructions, dependencies and check command.
- In Evals, choose New eval, upload the starter folder or enter its repository, then set the task and check command.
- Choose BoltzLabs models or your own endpoint. Each run gets its own machine.
- When the run finishes, open it to read the agent’s trace, file changes and check result.
Projects that need a desktop or phone, self-hosted websites, a game or a simulator need that runtime set up separately; an eval machine does not include it. Machine time uses account credits; hosted model calls are billed separately.