MCP connectors for AI video

By AI Video Atlas · License: Free, read-only

Our own free, read-only connector. Your agent can look up vendor-stated prices, sourced example shots and model guidance, and plan a shot list with a model, adapted prompt and cost estimate for each shot. It doesn't generate video itself.

Works with: Veo, Kling, Seedance, Hailuo, Runway, Luma, Wan, LTX, Sora, Grok Imagine

claude mcp add --transport http ai-video-atlas https://mcp.aivideoatlas.com/mcp

Official page ↗ · Checked 2026-10-16

By Runway · License: MIT · Last commit 2026-09-30 · ★ 23

Runway's own hosted connector. Sign in with your Runway account and generate images and video from ChatGPT, Claude, Cursor and other MCP apps, billed to your Runway credits. Runway archived its older self-hosted API server on GitHub and now points people to this hosted version.

Works with: Runway, Seedance, Kling, Veo

1. Go to chatgpt.com/plugins and search for Runway.
2. Click Install Plugin, then Sign in with Runway.
3. Mention @Runway in any chat and describe what you want.

The linked GitHub repo is the archived self-hosted server; license and last commit refer to it.

Repository ↗ · Official page ↗ · Checked 2026-10-16

By ElevenLabs · License: MIT · Last commit 2026-08-20 · ★ 1,535

ElevenLabs' hosted connector at api.elevenlabs.io/v1/mcp, with OAuth sign-in and nothing to install. It manages ElevenLabs voice agents and can generate speech audio from text, which is handy for voiceover. The older local server on GitHub is archived and deprecated in its favor.

Works with: ElevenLabs

https://api.elevenlabs.io/v1/mcp

The linked GitHub repo is the archived local server; license and last commit refer to it.

Repository ↗ · Official page ↗ · Checked 2026-10-16

By MiniMax · License: MIT · Last commit 2026-08-20 · ★ 1,590

MiniMax's official Python server. It gives an agent text-to-speech, voice cloning, image generation and Hailuo video generation (including Hailuo 02), with a separate tool to check on video jobs.

Works with: Hailuo

{
  "mcpServers": {
    "MiniMax": {
      "command": "uvx",
      "args": [
        "minimax-mcp",
        "-y"
      ],
      "env": {
        "MINIMAX_API_KEY": "insert-your-api-key-here",
        "MINIMAX_MCP_BASE_PATH": "local-output-dir-path, such as /User/xxx/Desktop",
        "MINIMAX_API_HOST": "api host, https://api.minimax.io | https://api.minimaxi.com",
        "MINIMAX_API_RESOURCE_MODE": "optional, [url|local], url is default, audio/image/video are downloaded locally or provided in URL format"
      }
    }
  }
}

Repository ↗ · Checked 2026-10-16

By MiniMax · License: MIT · Last commit 2026-08-21 · ★ 130

The official Node version of MiniMax's server, with the same image, Hailuo video, speech and voice cloning tools for people who'd rather run npx than Python.

Works with: Hailuo

{
  "mcpServers": {
    "minimax-mcp-js": {
      "command": "npx",
      "args": [
        "-y",
        "minimax-mcp-js"
      ],
      "env": {
        "MINIMAX_API_HOST": "<https://api.minimaxi.chat|https://api.minimax.chat>",
        "MINIMAX_API_KEY": "<your-api-key-here>",
        "MINIMAX_MCP_BASE_PATH": "<local-output-dir-path, such as /User/xxx/Desktop>"
      }
    }
  }
}

Snippet shortened: optional MINIMAX_RESOURCE_MODE line omitted; see the README.

Repository ↗ · Checked 2026-10-16

By PixVerse · License: MIT · Last commit 2025-10-22 · ★ 52

PixVerse's official server for text-to-video and image-to-video with its own models, with controls for quality, length and aspect ratio.

Works with: PixVerse

{
  "mcpServers": {
    "PixVerse": {
      "command": "uvx",
      "args": [
        "pixverse-mcp"
      ],
      "env": {
        "PIXVERSE_API_KEY": "your-api-key-here"
      }
    }
  }
}

Repository ↗ · Checked 2026-10-16

By Luma AI · License: MIT · Last commit 2025-04-17 · ★ 26

Luma's own small server for its Ray video models (Ray 2, Ray Flash 2, Ray 1.6) and Photon image models, with start and end keyframes, looping, and resolution up to 4K. It hasn't been updated since April 2025.

Works with: Luma

sh setup.sh

Run from a clone of the repo; the script asks for your Luma API key.

Repository ↗ · Checked 2026-10-16

By Google Cloud · Last commit 2026-10-16

A set of Go servers from Google Cloud's Creative Studio project for Veo 3 and 3.1 video, Gemini image (Nano Banana), Gemini TTS and Chirp speech, Lyria music, plus an audio and video compositing tool. You need a Google Cloud project. It lives in the repo's experiments folder.

Works with: Veo, Nano Banana, Lyria

curl -sL https://raw.githubusercontent.com/GoogleCloudPlatform/genmedia-creative-studio/main/experiments/mcp-genmedia/mcp-genmedia-go/install-online.sh | bash

Repository ↗ · Checked 2026-10-16

By Hugging Face · License: MIT · Last commit 2026-10-13 · ★ 304

Hugging Face's official hosted server. Besides searching models and papers, you can add Gradio Spaces as tools, including open video and image generators, from your Hugging Face MCP settings.

claude mcp add hf-mcp-server -t http https://huggingface.co/mcp?login

Repository ↗ · Official page ↗ · Checked 2026-10-16

By Luminary Lane (community) · License: MIT · Last commit 2026-05-04 · ★ 61

A community server for fal.ai that generates images, video (text-to-video and image-to-video), music and audio, using fal's queue for long jobs. It's not made by fal. It's the same project we list in guides as fal-mcp-server, now under a new GitHub owner.

Works with: fal, Kling, Flux

{
  "mcpServers": {
    "fal-ai": {
      "command": "uvx",
      "args": ["--from", "fal-mcp-server", "fal-mcp"],
      "env": {
        "FAL_KEY": "your-fal-api-key"
      }
    }
  }
}

Repository ↗ · Checked 2026-10-16

By deepfates (community) · License: MIT · Last commit 2025-09-24 · ★ 94

A community server for Replicate's API. Your agent can search and browse models, run predictions and check on them, so it can reach any video or image model hosted on Replicate. It's not made by Replicate.

Works with: Replicate

npm install -g mcp-replicate

Repository ↗ · Checked 2026-10-16

By artokun (community) · License: MIT · Last commit 2026-10-14 · ★ 829

A large community control layer for ComfyUI. An agent can generate images, video and audio, write and run workflows, and edit your live node graph in plain language, locally or on a server.

Works with: ComfyUI

{
  "mcpServers": {
    "comfyui": {
      "command": "npx",
      "args": ["-y", "comfyui-mcp"],
      "env": {
        "CIVITAI_API_TOKEN": ""
      }
    }
  }
}

Repository ↗ · Checked 2026-10-16

By joenorton (community) · License: Apache-2.0 · Last commit 2026-02-17 · ★ 410

A lightweight Python server that lets an agent run generations on a local ComfyUI install over streamable HTTP.

Works with: ComfyUI

git clone https://github.com/joenorton/comfyui-mcp-server.git
cd comfyui-mcp-server
pip install -r requirements.txt

Repository ↗ · Checked 2026-10-16

By ATH-MaaS (AIDC-AI) · License: MIT · Last commit 2025-12-17 · ★ 1,129

An open-source setup that turns ComfyUI workflows into MCP tools, so an agent can run your own image, video and audio workflows.

Works with: ComfyUI

uvx pixelle@latest

Repository ↗ · Checked 2026-10-16

By keugenek (community) · License: Apache-2.0 · Last commit 2026-03-29 · ★ 11

A community server for Krea.ai that reaches the video and image models Krea hosts, such as Hailuo, Kling, Runway, Veo, Pika and Flux, through one key. It's not made by Krea.

Works with: Krea, Hailuo, Kling, Runway, Veo, Pika, Flux

{
  "mcpServers": {
    "krea": {
      "command": "npx",
      "args": [
        "-y",
        "@vmosaic/krea-mcp-server"
      ],
      "env": {
        "KREA_API_KEY": "YOUR_KREA_API_KEY"
      }
    }
  }
}

Repository ↗ · Checked 2026-10-16

By vericontext · License: MIT · Last commit 2026-10-16 · ★ 178

A CLI and MCP server that turns a written brief into a plan, generates the clips with your own provider keys (Seedance, Runway, Kling and others) and renders a finished MP4. A hard spend cap means it can't run up your bill.

Works with: Seedance, Runway, Kling, Veo

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

Repository ↗ · Checked 2026-10-16

By MeiGen · License: MIT · Last commit 2026-10-15 · ★ 1,786

The connector for the MeiGen design service. Browse a prompt and inspiration library and generate images and short videos from ChatGPT, Claude or Cursor, with saved characters you can reuse. Generation uses MeiGen credits.

{
  "mcpServers": {
    "meigen": {
      "command": "npx",
      "args": ["-y", "meigen@3.2.0"]
    }
  }
}

Repository ↗ · Official page ↗ · Checked 2026-10-16