wtf is mcp-google-drive-agent-replit?

The Claude Asset Tagging Agent is a server that connects an AI assistant directly to Google Drive. When you give it a public Google Drive folder link, it reads the files inside, asks Claude to categorize each one based on its filename and metadata, and then writes those category tags back into the file's description field on Google Drive. It also adds a star to each processed file so you can filter and find tagged assets quickly. The project runs through two different server setups. The first is a local server for desktop use, communicating through standard input and output. The second is a web server designed for cloud hosting on a platform like Replit, which connects to Claude through a custom HTTP connector. Both servers talk to a shared backend built on Google Apps Script, which handles the actual interaction with Google Drive. This approach avoids the usual complex Google authentication process. Setting it up takes four steps. First, you paste the provided Google Apps Script code into a new project on Google's script editor and deploy it as a web app accessible to anyone. Second, you upload the project files to a Replit Node.js container, install dependencies, and start the server, which gives you a public URL. Third, you log into Claude, add that URL as a custom connector with "/mcp" appended to the end, and enable it. Fourth, you start a chat with Claude and ask it to list and tag the files in your Google Drive folder. The README includes a troubleshooting section covering common issues. These include a "Folder ID Not Permitted" error caused by Google Apps Script not pushing saved code live, connection timeouts from Replit's free containers going to sleep after inactivity, and bulk tagging failures when Claude sends too many parallel requests at once. Each issue comes with a specific fix, such as redeploying the script, retrying the prompt, or instructing Claude to process files one at a time rather than all at once. The repository is structured around four main files. Two are server files for local and remote use, one is the Google Apps Script backend, and one manages project dependencies. The README notes that this project serves as a portfolio demo and a practical lab for building AI agent tools.