web-agent-comparison

pleasedodisturb/web-agent-comparison

★ 0 stars Python AI/LLM Updated today

Reproducible 8-dimension benchmark of 7 browser-automation MCP servers (Playwright, Lightpanda, browser-use, Chrome DevTools, Firecrawl, Cloakbrowser, Obscura) against frozen loopback fixtures. v1.0.2 adds cross-platform evidence + real Chrome TLS fingerprint baseline. v1.1 expands fixtures to general web tasks.

View on GitHub → Try with Claude — $10 free →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "web-agent-comparison": {
      "command": "uvx",
      "args": [
        "web-agent-comparison"
      ]
    }
  }
}

Or install with pip: pip install web-agent-comparison

README Excerpt

**Which browser-automation MCP server should your AI agent use?** Vendors all claim "real Chrome." Cold-start latencies vary **51×** across the field. One leader ships as a closed-source binary that touches your cookies on launch. Real tradeoffs are hidden behind marketing copy. So we ran 7 of them on identical, frozen fixtures, scored every dimension that matters, and published the evidence. Pick a row.

Topics

agentic-browserai-agentsbenchmarkbrowser-automationclaude-codemcpmcp-serverplaywrightreproducibilityweb-scraping

web-agent-comparison

Quick Install

README Excerpt

Topics

Related AI/LLM Servers