Claude Code + Luma MCP: Generate AI Videos from a Single Command in the Terminal

📘 Complete Guide: Use All 11 AceData Cloud MCPs in Claude Code →

You’ve finished writing a README in Claude Code and want to add a dynamic demo. Opening a browser to use video tools breaks your flow—connect to the Luma MCP Server and get it done directly in the terminal without leaving the command line.

¶ Get API Token

Before using Luma MCP Server, you need an AceData Cloud API Token:

  1. Open AceData Cloud Platform, supports login via GitHub, Google, WeChat, email, or phone number. Register to get free credits.
  2. After logging in, you can see your API Token on the homepage—click to copy.
  3. When credits run out, top up at Console → Recharge, supports Alipay, WeChat, Stripe.

AceData Cloud Platform Homepage Token

One Token works for all 11 MCP Servers; no need to apply separately for each service.

💡 Using claude.ai web or Claude Desktop? They support OAuth one-click authorization, no manual token input needed. See Claude + Luma MCP Illustrated Tutorial for details.

¶ Configure Claude Code

Run this command in the terminal, replacing 䜠的Token with the API Token you copied:

claude mcp add luma --transport http https://luma.mcp.acedata.cloud/mcp \
  -H "Authorization: Bearer 䜠的Token"

⚠ -H must be uppercase. Lowercase -h is --help, not a header parameter.

Without -s, the default scope is local, which only applies to the current directory when running commands. It’s recommended to explicitly specify scope:

Scope Command Parameter Config File Scope of Effect
local (default) no -s or -s local ~/.claude.json Only the current project directory where the command runs
user -s user ~/.claude.json Global effect, available in all projects
project -s project .mcp.json in project root Current project, can be committed to git and shared with team

💡 To share the same config with your team, add -s project: the config is written to .mcp.json in the project root, which teammates can use after cloning from git. Do not commit real Tokens to public repos; use ${ENV_VAR} placeholders plus environment variables.

Verify it’s working: run claude mcp list, you should see luma showing ✓ Connected.

¶ Practical Scenarios

After configuration, return to your Claude Code session and call Luma directly with natural language:

Create a promo video after finishing a project

Post it on Twitter or Product Hunt:

Generate a video: a laptop screen showing code, a steaming cup of coffee beside it, soft morning light

Animate design mockups

luma_generate_video_with_reference accepts images or videos as references to turn static screenshots into animated demos:

Generate a video using this UI screenshot as reference, adding subtle animations to the interface elements

Video too short? Continue

luma_extend_video continues from the end, keeping the style consistent:

Extend this video by 10 seconds, maintaining the same style

¶ Tool List

Tool Description
luma_generate_video Text-to-video generation
luma_extend_video Video extension
luma_generate_video_with_reference Reference image/video based generation