Bingeable

MCP

Make product videos from any AI agent

Connect Claude, ChatGPT, Gemini, Cursor, or your own support bot to Bingeable. Ask “how do I invite a teammate?”, reuse an existing walkthrough if you have one, or generate a new video and get back a public embed link.

Generate your first video for freeManage connected apps

How it works

Ask in the agent. Approve the wording. Share the link.

Connect once

Claude, ChatGPT, Gemini, or Cursor signs in with OAuth. After that, the agent can search your library and prepare videos.

Approve the prompt

The agent rewrites the request to Bingeable’s prompt guidelines and shows you that wording. Nothing records until you say yes.

Share the embed

When the video is ready, the agent returns the public embed link. Paste it into a help article, a doc, or a reply.

Use cases

Where teams ask an agent for the video

Support reply

“How do I invite a teammate?”

Search the library first. Reuse an existing walkthrough, or approve a new one and send the embed with the written answer.

Docs from Cursor

“Walk through connecting Stripe from Settings.”

The rewritten prompt is approved in chat, then the embed lands in the doc or the pull request.

Launch video

“Make a showcase of our pricing page.”

ChatGPT or Gemini can brief a public-page showcase. No login. Approve the wording, then share the link.

Content gaps

“Which questions still have no video?”

Questions the library could not answer are logged. Review the backlog and approve the smallest set of videos that covers them.

Remote connector

Claude, ChatGPT, and other remote MCP clients connect over HTTPS. Sign in once, approve the scopes, and the agent can search your library. A new video starts only after you approve the rewritten prompt.

MCP server URL

https://bingeable.ai/api/mcp

Claude Desktop / remote MCP config

{
  "mcpServers": {
    "bingeable": {
      "url": "https://bingeable.ai/api/mcp"
    }
  }
}

Cursor and CLI

Clients that only speak stdio can run the thin wrapper. It forwards to the same hosted server and uses the same OAuth login in the browser.

Cursor mcp.json

{
  "mcpServers": {
    "bingeable": {
      "command": "npx",
      "args": ["-y", "@bingeable/mcp"]
    }
  }
}

What an agent can do

  • find_or_create_video — the support-agent primitive. Search first. A new video is rewritten to the prompt guidelines and starts only after the person approves that wording.
  • get_embed_code — returns the public /embed/{id} link and iframe snippet.
  • get_job_status — poll a render. Videos take minutes, not seconds; the response tells the agent how long to wait.
  • record_video_request — log a question the library could not answer, so you can see the content gap.