YouTube

What Klavis-AI/YouTube MCP server does

Klavis-AI/YouTube MCP server provides an MCP interface for extracting and converting YouTube video information. The supplied tool list shows a browser-oriented implementation: an agent can open a URL, inspect the page, interact with its controls, and retrieve page or network details.

The server is therefore suited to workflows where the required information is available through a YouTube web page. It does not expose a small set of named YouTube operations in the provided material, such as a dedicated video-search or transcript tool. Instead, the agent works with general browser capabilities.

How it works

Klavis-AI/YouTube MCP server uses browser actions to operate on a current page. Navigation and history tools move between URLs, while snapshots and text search help an agent locate content in the page's accessibility tree. Clicking, typing, form filling, selecting options, hovering, dragging, keyboard input, and dialog handling support interaction with page controls.

For inspection, the server can return browser console messages, list network requests, and retrieve full details for an individual request. It can also capture screenshots, wait for text to appear or disappear, resize the browser window, and manage multiple tabs. JavaScript can be evaluated against a page or element. The browser_run_code_unsafe capability runs Playwright code in the server process and is explicitly described as RCE-equivalent, so it should be treated as a high-risk operation.

Setup and configuration

The repository README documents several ways to use Klavis services: a cloud-hosted option at klavis.ai, self-hosted Docker examples for another Klavis MCP server, and SDK or REST API examples for creating server instances. The supplied material does not provide a YouTube-specific Docker image, package name, startup command, MCP configuration block, or required environment variable.

Klavis-AI/YouTube MCP server can therefore be evaluated as part of the Klavis MCP integration offering, but a developer should consult the linked Klavis documentation for the exact deployment path for this server. The README's SDK and REST examples use a Klavis API key for those platform-management operations; the material does not establish that the local YouTube browser server itself requires that credential.

Tools and capabilities

Available capabilities include:

• Navigate to URLs and return to earlier pages.
• Capture accessibility snapshots, screenshots, console messages, and network activity.
• Search page snapshots for text or regular expressions.
• Click, type, fill forms, select options, hover, drag, upload files, and handle dialogs.
• Create, close, list, and select browser tabs.
• Wait for page text or a specified period.
• Evaluate JavaScript on a page or element.
• Run Playwright code through the unsafe execution tool.

These tools support browser-level extraction and conversion workflows, but the supplied material does not specify the output format or identify which video fields are converted.

Limitations and notes

The provided description is limited to extracting and converting YouTube video information, while the detailed tool list describes general browser automation. No YouTube API credentials, rate limits, supported video fields, transcript behavior, or conversion formats are documented in the supplied material.

Klavis-AI/YouTube MCP server also has no project-specific installation command stated in the excerpt. Compatibility with Claude Desktop, Cursor, Windsurf, and Cline is not confirmed. Treat arbitrary-code execution as sensitive because the README identifies the unsafe Playwright tool as equivalent to remote code execution in the server process.