Treza MCP server

Treza plugin

Agent Plugins package for the Treza MCP server. Install it in Cursor, Codex, GitHub Copilot, VS Code, or any client that supports the standard, and your agent can build, run, schedule, and publish AI video pipelines to YouTube and TikTok. The same repo is a Gemini CLI extension, and Cline can install it from llms-install.md.

• Server: https://www.trezalabs.com/api/mcp (Streamable HTTP, OAuth with dynamic client registration)
• Landing page and setup snippets: https://www.trezalabs.com/connect
• Docs: https://docs.trezalabs.com/api/mcp-server
• Privacy: https://docs.trezalabs.com/terms-and-privacy/privacy

Two ways to connect

Remote (recommended). Point any client that supports OAuth at https://www.trezalabs.com/api/mcp. The first connection signs you in with Google and creates the Treza account. Nothing to install.

stdio bridge. For clients and agent runtimes that only spawn local servers, this repo ships treza-mcp, a small Node bridge that forwards every tool, prompt, and resource to the hosted endpoint. It needs a scoped API key in TREZAAPIKEY (mint one in the Treza app under Settings, API keys, or have an OAuth-connected client call createapikey).

{
  "mcpServers": {
    "treza": {
      "command": "npx",
      "args": ["-y", "treza-mcp"],
      "env": { "TREZA_API_KEY": "treza_live_..." }
    }
  }
}

TREZAMCPURL overrides the endpoint (staging, self-hosted). The bridge runs nothing locally; it is a proxy.

What is inside

Tools

Twenty-four tools in five groups:

• Authoring: listpipelinetemplates, listnodetypes, getnodetype, createpipeline, updatepipeline, publish_pipeline
• Running: listpipelines, getpipeline, runpipeline, getrun, listruns, setschedule_paused
• Your media: listassets browses the media library, importmedia adds a file (an attachment, a public link, or a local file sent through createuploadurl), assemblevideo joins clips into one video with optional narration, a music bed, and burned-in captions, and editasset runs node operations (captions, upscale, trim, crop, and more) over a single file
• Channels and publishing: listconnectedchannels, connectchannel returns a link that connects a YouTube channel or TikTok account in the person's own browser, and publishasset lays a finished video out for review before it posts. confirm_publish belongs to the publish panel: it runs when the person presses Post there, never from the model
• Budget and access: estimateruncost, getcreditbalance, createapikey (OAuth connections only)

Read-only tools carry readOnlyHint. runpipeline, publishpipeline, assemblevideo, and editasset spend prepaid credits, and runpipeline and publishpipeline can post to a connected channel. assemblevideo and editasset render in the background and return a run to poll with get_run.

In clients that show MCP Apps panels, such as ChatGPT and Claude, results appear as panels: the finished file, the price with a Start render button, a library to pick files from, a map of a pipeline, and the post before it goes out. Other clients get the same answers as text. Posting with publish_asset needs a panel; without one, publish through a pipeline that ends in a youtube-upload or tiktok-upload node.

Shortened. The full README is on GitHub.