What is an MCP server?
The Model Context Protocol (MCP) is an open standard that lets AI assistants call outside tools and services. An MCP server is the piece a company runs so that assistants can use its product. When you add a server to Claude, ChatGPT or Cursor, the assistant gains a list of tools it can call during a conversation, and it decides which ones a request needs.
What makes a video MCP server useful?
Video work is slow, multi-step and easy to describe in words: "dub this into Spanish", "cut three shorts", "make a thumbnail that matches my channel". That makes it a natural fit for an agent. The Braiv MCP server exposes each step as a tool, tracks the long-running jobs behind them, and returns the results to the conversation, so one sentence can replace an afternoon of exporting and re-uploading.
How the Braiv MCP server works
The server is hosted by Braiv at https://mcp.braiv.co/mcp. Your AI tool connects over Streamable HTTP and signs you in with OAuth, the same account you use in Braiv Studio. Each tool wraps an operation in the Braiv API, so anything an agent does over MCP follows the same rules, plan limits and credit pricing as the API and the app.
Jobs, not waiting
Importing, dubbing and clipping take time, so every long-running action returns a job. The agent checks the job until it is ready, then carries on. You can close the chat and the work still finishes in your Braiv library.
Privacy and control
Your assistant receives an access token, never your password. Signed download links are hidden from tool output so they are not echoed into a chat. You can disconnect Braiv from your AI tool whenever you like.