Agent-ready — Model Context Protocol

Connect your agentic clipper over MCP

LumiClip is an autonomous AI video clipper — and it speaks MCP. Add the LumiClip server to Claude, Cursor, or any MCP client and let your agent turn YouTube and Twitch links and channels into captioned vertical clips, then publish them, as tool calls.

Connect it REST API docs

What your agent can do

Every step of LumiClip's pipeline is exposed as an MCP tool, so the model can run the whole thing unattended.

Watch channels

Ask the agent to monitor up to 10 Twitch streamers or YouTube channels. LumiClip detects new uploads and kicks off clipping without another prompt.

Clip + caption

The agent submits a YouTube link (or channel) and gets back AI-ranked vertical clips with burned-in captions and 9:16 speaker tracking — as structured tool results.

Publish

Auto-publish the finished shorts to TikTok, Instagram Reels, Facebook, and YouTube Shorts, or hand the download URLs back to your own workflow.

Compose in code

Because it's MCP, LumiClip drops into any agent framework or multi-tool chain — combine it with your other connectors and let the model orchestrate the whole pipeline.

Connect in three ways

Install from the Smithery directory, use the hosted remote endpoint, or run the server locally via npm. Each registers LumiClip as a set of tools your MCP client can call.

Smithery directory

One-click install from our listing in the Smithery MCP registry:

View on Smithery

Hosted remote (fastest)

Add a custom connector pointing at our hosted endpoint:

https://mcp.lumiclip.ai/mcp

Local (npm package)

Point your MCP client's config at the package:

{
  "mcpServers": {
    "lumiclip": {
      "command": "npx",
      "args": ["-y", "@lumiclip/mcp-server"],
      "env": { "LUMICLIP_API_KEY": "your_key" }
    }
  }
}

Why an agent, not just a clipper

  • Set it up once — point it at the channels you follow and it keeps clipping new uploads on its own.
  • No timeline to babysit: detection, ranking, captioning, and publishing all run unattended.
  • Callable by other agents and workflows via MCP or the REST API — LumiClip becomes one tool in a larger chain.
  • Still fully manual when you want it — import any video and edit every clip by hand.

MCP FAQ

What is the LumiClip MCP server?

It's a Model Context Protocol (MCP) server that exposes LumiClip's agentic clipping pipeline as tools an AI assistant can call. Connect it to Claude, Cursor, or any MCP client and the model can turn a YouTube link or channel into captioned vertical clips, then publish them — all as tool calls, no manual timeline editing.

Does LumiClip work with Claude?

Yes. Add the LumiClip MCP server to Claude Desktop (or any MCP-compatible client) as a custom connector using the hosted endpoint at https://mcp.lumiclip.ai/mcp or the @lumiclip/mcp-server package. Claude can then submit videos and channels and fetch finished clips directly in a conversation.

Is LumiClip listed in Claude's Connectors Directory?

Not yet — a submission to Anthropic's official Connectors Directory is planned. In the meantime you can connect LumiClip today as a custom MCP connector using the endpoint or npm package on this page. (Note: our older Smithery listing is a third-party MCP directory, not Anthropic's own connector list.)

What can an agent do through the MCP server?

Submit a YouTube link or channel, watch channels for new uploads, generate AI-ranked vertical clips with captions and 9:16 reframing, retrieve download URLs and metadata, and auto-publish to connected social accounts — all as MCP tool calls.

Add LumiClip to your agent

Grab the server from the Smithery MCP directory, or start free and generate an API key to connect it yourself.

View on Smithery Start free