Today I ran an essential tool for Vibe Coding: Chrome DevTools MCP, which already has 17.6K stars on GitHub. When using Vibe, we often need to obtain external data, such as the latest trending tokens on GMGN or Twitter tweets. Traditional web scraping is no longer suitable for the AI era. As an MCP, Chrome DevTools is very easy to use, requiring only two steps and three minutes. First, install MCP; second, when interacting with the AI, simply tell it to use MCP. Chrome DevTools basically covers all Vibe Coding tools. I personally used it on Antigravity and Claude Code. Unfortunately, Antigravity didn't work properly with different models; the agent would run for a while and then try to install other traditional web scraping libraries. Claude Code, however, worked very easily and successfully. Chrome DevTools automatically opens the Chrome browser to begin working. One of its biggest advantages is that users can open the browser and log in beforehand. This allows the AI to manage already logged-in websites. Before using Claude Code's "niuma" (or similar tool) to work, it's recommended to check if MCP is installed correctly using `/mcp`. My first two attempts failed, and I initially thought it was a problem with the "niuma" itself, but it turned out MCP wasn't configured correctly. My task this time was to compile a list of KOLs on the Yapper leaderboard on Kaito, hoping to further utilize this information to track their tweets. I first logged into Kaito in Chrome and then let the "niuma" tool work. The "niuma" tool is very intelligent; it only requires clear explanations in plain language, making it very simple and requiring no complicated prompts. It gave me the results in about 5 minutes. One more thing to note: if you need to use a browser while the "niuma" tool is working, it's recommended to configure a new Chrome browser and place it on a virtual desktop.