MCP video generator

MCP video generator: one server for every AI client

EveryGen is a remote MCP server over Streamable HTTP. Any client that supports it can list video models, start a generation and fetch the result. Codex, Claude and your own agents all use the same address and the same credits.

  1. 1

    Add the server to Codex

    Open a terminal and run this once. It registers EveryGen as a remote MCP server in your Codex configuration:

    codex mcp add everygen --url https://everygenai.video/api/mcp
  2. 2

    Sign in

    Run this command. Your browser opens the everygenai.video sign-in page: log in with the account whose credits you want to use, click Allow and go back to the terminal:

    codex mcp login everygen
  3. 3

    Other clients: use the server URL

    In the client's MCP settings add a remote server of type HTTP (Streamable HTTP) and paste this address. Clients with OAuth open a sign-in window like Claude does. Others can send an EveryGen API key, created in your account settings, as an Authorization: Bearer header:

    https://everygenai.video/api/mcp
  4. 4

    Ask for a video

    Describe the clip in your client and approve the tool call if it asks. The client calls list_video_models to pick a model, generate_video to start the job and get_video_task to fetch the finished video, which usually takes a minute or two.

AI assistant + EveryGen
Make a 5-second vertical clip of a golden retriever surfing at sunset.
EveryGen · generate_video

Done. Seedance 2.0 Mini, 9:16, 720p, 5 seconds. Here is your video.

Questions

What is an MCP video generator?

A Model Context Protocol server that exposes video generation as tools. An AI assistant or coding agent calls them to create a video, so you can make clips without leaving the client.

Which clients work?

Any client that supports remote MCP over Streamable HTTP, including Claude, Claude Code and Codex. Clients with OAuth sign in directly; others use an API key as a bearer token.

Which tools does it expose?

list_video_models, generate_video and get_video_task: list the models and prices, start a text-to-video or image-to-video job, and fetch the finished video.

How is it billed?

Per video from the credits in your EveryGen account, the same as on the website. Failed jobs are refunded.

What the connector can do

  • list_video_models

    Lists every video model with its durations, resolutions, aspect ratios and credit cost per second.

  • generate_video

    Starts a text-to-video or image-to-video job. Uses credits from the connected account; failed jobs are refunded.

  • get_video_task

    Checks a job and returns the video link when it is ready.

Ask in plain words, get a finished clip

Real results from the connector. Each clip came from the request on the left; the assistant turned it into the full prompt shown under the video.

Make a 5-second vertical clip of a golden retriever surfing at sunset.

Prompt sent to EveryGen

A golden retriever riding a small surfboard on a gentle wave at sunset, slow motion spray, warm golden light, joyful, cinematic, no text

I need a premium product shot for a perfume launch. Vertical, no text.

Prompt sent to EveryGen

Luxury product shot: a glass perfume bottle standing on wet black stone, water droplets sliding down, slow orbiting camera, soft studio rim light, premium commercial look, no text or logos

Give me a moody drone shot over a rainy neon city for a YouTube intro.

Prompt sent to EveryGen

Cinematic drone shot gliding over a rainy neon-lit city street at night, reflections on wet asphalt, umbrellas moving below, moody blue and pink light, no text

Something cozy for a story: a paper boat sailing across a desk.

Prompt sent to EveryGen

Stop-motion style: a small paper boat sails across a wooden desk past a coffee cup and pencils, warm desk-lamp light, playful and cozy, no text

Your next video is one message away

Create your EveryGen account, connect it to Claude or ChatGPT, then ask for your first clip.