Yutu

What eat-pray-ai/yutu MCP server does

eat-pray-ai/yutu MCP server provides a CLI and MCP interface for automating YouTube channel operations. The documented workflow covers uploading and optimizing videos, managing comments and playlists, and handling channel branding. It can also retrieve, create, update, and delete YouTube content.

The repository includes a separate agent mode, called Miffy, for automating YouTube workflows. That mode includes YouTube growth strategy and SEO capabilities according to the project documentation, but it is distinct from the core CLI and MCP server usage.

How it works

The application uses a Google Cloud project and YouTube APIs. YouTube Data API v3 is required. YouTube Analytics API and YouTube Reporting API can be enabled when analytics or reporting access is needed; both are listed as optional prerequisites.

Authentication uses an OAuth desktop-app client created in Google Cloud. Running the authentication command opens a browser for consent and saves a token locally. Subsequent commands read the client secret and cached token from the current directory by default. The paths can instead be supplied through environment variables or the authentication command's credential and token flags.

The agent mode accepts a Gemini API key and can run in console mode. The README also documents web-oriented launch arguments and model selection. Agent mode is currently limited to Google's Gemini models.

Setup and configuration

Create a Google Cloud project, enable the required YouTube API, configure an OAuth consent screen, and add yourself as a test user. Create a Desktop app OAuth client and save its downloaded credentials as client_secret.json.

Install the CLI using one of the documented distribution methods, such as the npm package, Homebrew, Go, Docker, a Linux installation script, or WinGet. The npm package is @eat-pray-ai/yutu.

Authenticate with:

yutu auth --credential client_secret.json

This stores the OAuth token as youtube.token.json by default. Configuration variables include:

• YUTU_CREDENTIAL: OAuth client secret path, Base64 value, or JSON content.
• YUTU_CACHE_TOKEN: cached OAuth token path, Base64 value, or JSON content.
• YUTU_ROOT: root directory used for resolving files.
• YUTU_LOG_LEVEL: logging level, such as DEBUG, INFO, WARN, or ERROR.

Tools and capabilities

The documented YouTube operations include:

• Uploading videos.
• Optimizing video content.
• Retrieving YouTube content.
• Creating, updating, and deleting content.
• Managing comments and playlists.
• Managing channel branding.

The eat-pray-ai/yutu MCP server is intended for MCP-based automation, while the same project also exposes a direct CLI. The repository additionally documents an AI agent mode for higher-level workflow automation.

Limitations and notes

A Google Cloud account and OAuth setup are required. YouTube Data API v3 must be enabled before use. Analytics and Reporting APIs are optional rather than prerequisites for the basic YouTube workflow.

The README excerpt does not identify specific MCP client configurations or list individual MCP tool names. It also does not establish whether YouTube API usage may incur Google Cloud charges, so pricing should be checked against Google's current terms. Agent mode requires a Gemini API key and currently supports only Google's Gemini models.