Repositorynekuda-ai/WindTunnelResearch
Standard browser agents rely heavily on expensive visual parsing or complex DOM navigation. WindTunnel provides concrete empirical data evaluating whether standardized, site-exposed actions via WebMCP can offer a significantly faster and cheaper alternative for web-based agent workflows.
WindTunnel tackles the operational friction of web automation by providing comparative data across UI manipulation approaches. It targets AI researchers, web automation engineers, and developers designing agent interfaces. The tool is implemented using Node.js, Playwright, and containerised Linux environments. In terms of maturity, while community adoption is early (34 stars), the codebase exhibits robust testing discipline, including a zero-cost local test harness, comprehensive CI workflows, and a heavily documented test run consisting of thousands of evaluated attempts.
Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.
Similar in content, found by the AI search.
Link:nekuda-ai
Nekuda-ai has released WindTunnel, a reproducible open-source benchmark evaluating WebMCP against screenshot-based computer use and DOM-based browser agents. Evaluating 49 tasks across eight real websites with three attempts per task, the benchmark tests models including GPT-5.6 Luna, Gemini 3.6 Flash, and Sonnet 5.
X post:Idan Levin
Idan Levin introduced WindTunnel, an open benchmark comparing WebMCP tool calling against computer-use and DOM-based approaches across 49 tasks on eight real websites.
581bookmarks138.3kviews
Repository:ChromeDevTools/chrome-devtools-mcp
chrome-devtools-mcp connects AI coding agents to Google Chrome via the Model Context Protocol (MCP). It enables agents to automate browser actions using Puppeteer, inspect network traffic and console errors with source maps, and collect detailed performance traces.
50.8kstarsTypeScript
Repository:Podcast72/showroom-la-finestra-webmcp
The showroom-la-finestra-webmcp repository demonstrates how existing websites can expose structured business knowledge and safe actions directly to AI agents without running a separate MCP server. By using an in-browser layer via `document.modelContext`, the website explains itself rather than relying on DOM scraping.
1starsJavaScript
Repository:per-simmons/simmonsbench-agent-fanout
agent-fanout spawns up to six parallel Claude Code instances in iTerm2, having them build tasks simultaneously while collecting token counts, costs, and runtimes for model comparisons.
4starsShell
Repository:captivus/chrome-agent
chrome-agent is a CLI tool and Python library giving AI coding agents direct browser inspection and interaction via the Chrome DevTools Protocol (CDP). It avoids restrictive abstraction layers and enables concurrent, isolated use by multiple agents and human operators.
243starsPython
Parts of this content were created with the assistance of AI.
