Skip to content

Update: ClawBench entry — V2 corpus (283 tasks, 163 platforms) #60

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
nibzard merged 1 commit into from
May 26, 2026
Merged

Update: ClawBench entry — V2 corpus (283 tasks, 163 platforms) #60

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -189,7 +189,7 @@ Datasets, benchmarks, and notable research efforts for evaluating and advancing
- [WorkArena](https://github.com/ServiceNow/WorkArena) - A suite of 33 browser-based tasks for enterprise "knowledge worker" scenarios. ![GitHub Repo stars](https://img.shields.io/github/stars/ServiceNow/WorkArena?style=social)
- [BrowserGym by ServiceNow](https://github.com/ServiceNow/BrowserGym) - A gym environment for web task automation. ![GitHub Repo stars](https://img.shields.io/github/stars/ServiceNow/BrowserGym?style=social)
- [TimeWarp](https://timewarp-web.github.io) - A benchmark on historical versions of web UI.
- [ClawBench](https://github.com/reacher-z/ClawBench) - 153 everyday tasks on 144 live production websites across 15 categories, with a submission-interception layer that blocks only the final write request to preserve real-site behavior without side effects. ![GitHub Repo stars](https://img.shields.io/github/stars/reacher-z/ClawBench?style=social)
- [ClawBench](https://github.com/reacher-z/ClawBench) - 283 everyday tasks (V1 153 + V2 130) on 163 live production websites across 15 categories. Two-stage scoring (final HTTP-request interception + LLM judge) blocks only the write request so real sites stay clean. Public leaderboard with 5-layer execution traces (recording, action log, request log, agent messages, interception). ![GitHub Repo stars](https://img.shields.io/github/stars/reacher-z/ClawBench?style=social)

## Tutorials & Guides

Expand Down
Loading