Skip to main content
browseruse-bench provides a unified interface to run and evaluate different browser agents.

​
Supported Agents

browser-use

Python-based browser agent with programmable browser automation capabilities. Supports both local browsers and Lexmount cloud browsers.

Agent-TARS

Browser agent developed by the TARS team, distributed as an npm CLI package.

Skyvern

Browser automation agent powered by the Skyvern SDK with multiple engines and browser backends.

Claude Code

Anthropic’s official Claude CLI, driving a real browser through Playwright MCP in non-interactive mode.

​
Feature Comparison

Featurebrowser-useAgent-TARSSkyvernClaude Code
LanguagePythonNode.jsPythonNode.js CLI
Installationuv/pipnpmuv/pipnpm
Local Browser✅✅✅✅ (via MCP)
Cloud Browser✅✅✅—

​
Quick Start

# Install
uv sync

# Run
bubench run --agent browser-use --benchmark LexBench-Browser --split L1 --mode first_n --count 3
If bubench is not found, activate .venv or run commands via uv run bubench ....

​
Planned Support

  • More open-source agents
If you’d like to integrate your own agent into browseruse-bench, please refer to the Contributing Guide.