GITHUB REPOSITORY
ClawBench Browser-Agent Evaluation
tiger-ai-lab/clawbench
ClawBench evaluates browser agents on live-site tasks with recorded traces and interception-based scoring. Its implemented Hermes harness attaches native browser tools to the benchmark's instrumented Chromium session.
Why it's here
Primary evidence supports the described AI role; AI authorship of the implementation is not established.
★ 1.0k starsLast commit Oct 13, 2026Chrome DevTools ProtocolChromiumDockerHermes AgentPodman
Stars and last commit as of Oct 14, 2026. Image is GitHub's social preview for this repo.