browser-tool-benchmark
Find out whether Playwright CLI or MCP is cheaper for your agent
Run the published benchmark on your machine and compare tokens, cost, speed, and pass rate with the published results.
When to use it
- You want to compare Playwright CLI and MCP on your model.
- You need token and cost measurements for the same browser tasks.
- You want to reproduce ThinkRun’s published benchmark results.
Supported agents
- Claude Code
- Codex
- Cursor
- Gemini CLI
The repository mirrors the same skill package in each agent’s native directory.
Install and inspect the source
Clone the public ThinkRun repository, then install this skill in your agent’s native skills directory.
mkdir -p .claude/skills/browser-tool-benchmark && curl -fL https://thinkrun.ai/skills-download/browser-tool-benchmark/SKILL.md -o .claude/skills/browser-tool-benchmark/SKILL.md