WebsiteBench / README.md
AgPerry's picture
Add WebsiteBench card, manifest, and the medium task
bc7aa4f verified
|
Raw History Blame
3.86 kB
---
license: apache-2.0
pretty_name: WebsiteBench
language:
- en
tags:
- benchmark
- agents
- web-agents
- website-reconstruction
- harbor
viewer: false
extra_gated_heading: Access WebsiteBench
extra_gated_description: >-
The reference websites in WebsiteBench reproduce the appearance and behavior of
real websites. Their text, images, fonts, and trademarks belong to their owners.
extra_gated_prompt: >-
By requesting access you agree to use WebsiteBench for non-commercial research
only and not to redistribute the reference website content outside this dataset.
extra_gated_fields:
I will use WebsiteBench for research only: checkbox
I understand that website content belongs to its owners: checkbox
extra_gated_button_content: Agree and access
---
# WebsiteBench
Given a production-level website they can explore through a browser, agents must
reverse-engineer a full-stack version that matches its appearance and behavior as
closely as possible.
This dataset holds the 70 WebsiteBench tasks in the native task format of the
[Harbor framework](https://github.com/harbor-framework/harbor) (version 0.25.0).
The same task contents are published on [Harbor Hub](https://hub.harborframework.com/)
as `reacher-z/WebsiteBench@v1`; every task has the same Harbor content hash in both
places. Code, documentation, and the evaluation commands live on
[GitHub](https://github.com/TIGER-AI-Lab/WebsiteBench).
- Project page: [website-bench.com](https://website-bench.com/)
- Code: [TIGER-AI-Lab/WebsiteBench](https://github.com/TIGER-AI-Lab/WebsiteBench)
## Download and run
Accept the access terms on this page, log in with `hf auth login`, then use the
download script from the GitHub repository. It verifies every file against
`manifest.json` and restores the executable permissions that Hugging Face does not
store:
```sh
git clone https://github.com/TIGER-AI-Lab/WebsiteBench.git
cd WebsiteBench
uv sync
uv run python scripts/download_tasks.py --source hf --task medium
uv run harbor run --path ./local-tasks/medium --agent oracle
```
Omit `--task` to download all 70 tasks into `local-tasks/`. The GitHub
[documentation](https://github.com/TIGER-AI-Lab/WebsiteBench/tree/main/docs)
describes coding-agent runs, batch runs, and results.
## Contents
| Path | Contents |
| --- | --- |
| `tasks/<task>/instruction.md` | The reconstruction task given to the agent |
| `tasks/<task>/task.toml` | Harbor configuration, resources, and network policies |
| `tasks/<task>/environment/` | Agent environment, browser gateway, mailbox, and the reference website |
| `tasks/<task>/solution/` | The bundled oracle implementation installed by `solve.sh` |
| `tasks/<task>/tests/` | The separate verifier, its oracle runtime, and the hidden test suite |
| `manifest.json` | SHA-256, size, and mode of every file, and each task's Harbor content hash |
## Validation
During benchmark construction, every task's bundled oracle solution was run through
its verifier; these runs are historical records. The files in this release were
checked statically: each task passes the repository's asset check, and every file
matches `manifest.json` and the Harbor Hub package.
## License and terms
The benchmark code, task definitions, and verifier are released under the
Apache-2.0 license. The reference websites reproduce third-party sites for
evaluation; their content remains the property of its owners and is provided for
non-commercial research only, as stated in the access terms.
Each task file carries a `harbor-canary` GUID. Please exclude documents containing
these strings from training data.
## Citation
```bibtex
@misc{zhang2026websitebench,
title = {WebsiteBench: Can AI Agents Rebuild Websites through Browser Exploration?},
author = {{WebsiteBench Authors}},
year = {2026},
url = {https://github.com/TIGER-AI-Lab/WebsiteBench},
}
```