Let an AI agent like Claude make videos on your account, or call the REST API from your own code.
Add NextGen Studio to Claude or ChatGPT with this address, then sign in. No key or setup file needed.
/mcp and choose Authenticate.claude mcp add --transport http nextgen-studio https://nextgenstudioai.com/mcpA NextGen Studio page opens. Sign in, check what the app will be able to do, and click Allow.
In Claude and ChatGPT you also get galleries to click for art styles, voices and subtitle styles, a live progress card, and a player for the finished video.
Paste this into Claude, ChatGPT, Cursor or your own agent. It tells your AI how to connect and how to work with you: show you the cost first, stop so you can check the script, and ask before spending credits.
You can make videos for me with NextGen Studio, an AI video maker. You work on my account and spend my credits, so follow these rules. How to connect: - If you can use MCP servers, add https://nextgenstudioai.com/mcp as a remote server (HTTP). When it asks, I'll sign in with my NextGen Studio account. - If you can't, call the REST API at https://nextgenstudioai.com/v1 with the header Authorization: Bearer followed by my API key. I'll give you the key. Keep it in an environment variable, never in code, files or messages you share. - Every request and tool is listed on the Developers page: https://nextgenstudioai.com/developers How to work with me: 1. Start by checking my plan and credits (get_account or GET /v1/account) and the choices (list_options or GET /v1/options). 2. If I haven't picked an art style, a voice or a subtitle style, show me the options (show_styles, show_voices, show_caption_styles) and let me choose. 3. Tell me what the video will cost (estimate_cost or POST /v1/estimate) and wait for my OK before you create it. 4. Create it with autopilot set to false unless I ask you to run it all the way. It then stops after the script and after the scene pictures so I can check them. Continue (continue_video) only when I say so. 5. While it's being made, call wait_for_video (GET /v1/videos/{id}/wait) instead of checking in a fast loop, and show me its progress line each time. 6. When it's done, give me the download link (get_download_link). If it's for YouTube, also get the upload package (get_publish_kit). 7. Ask me before anything else that spends credits (a new scene picture, animating a scene, rendering again, a thumbnail) and before you delete a video.
Once it's connected, talk to your AI in plain words. Copy one of these to start.
What art styles does NextGen Studio have? Show me the pictures.
Show me the NextGen Studio voices in English so I can listen to a few and pick one.
Make a 5-minute video with NextGen Studio about the history of coffee, in style 3, with a calm male voice and Karaoke Pop subtitles. Tell me the cost first.
Make a 30-second Short with NextGen Studio about why octopuses have three hearts, with a few scenes animated.
Make a NextGen Studio video from my script below, read exactly as written: [paste your script]
Make a NextGen Studio video about [your topic], but stop after the script so I can check it.
Redo the picture for scene 4 as a close-up of coffee beans, then render the video again.
Give me the download link for my latest video.
Give me the YouTube upload package for my latest video, and make a thumbnail for it.
I've downloaded my coffee video. Delete it so I have room for new ones.
Each one acts on your account and spends your credits. Apps you connected by signing in show here too.
Loading your keys…
For scripts and servers that can't sign in. Run this once in your terminal.
claude mcp add --transport http nextgen-studio https://nextgenstudioai.com/mcp --header "Authorization: Bearer ngs_live_YOUR_KEY"In Claude Desktop, open Settings, then Developer, then Edit Config. Add this to claude_desktop_config.json and restart the app.
{
"mcpServers": {
"nextgen-studio": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://nextgenstudioai.com/mcp", "--header", "Authorization:${NGS_AUTH}"],
"env": { "NGS_AUTH": "Bearer ngs_live_YOUR_KEY" }
}
}
}This setup runs through npx, so it needs Node.js installed.
Any MCP client that supports remote servers over HTTP can connect. Your own code can call the same actions over REST.
Try it with curl:
curl -H "Authorization: Bearer ngs_live_YOUR_KEY" https://nextgenstudioai.com/v1/accountEach request has a matching MCP tool, so an agent and your own code can do the same things.
Add webhook_url when you create a video, and we send a POST to it when the video finishes, fails, is cancelled or pauses for your review.
{
"topic": "Why octopuses have three hearts",
"mode": "quick_take",
"webhook_url": "https://example.com/nextgen-hook"
}Each request carries an NGS-Signature header: an HMAC-SHA256 of the timestamp and the body, made with your key's webhook_secret from GET /v1/account.
import hashlib, hmac, time
def verify(headers, body: bytes, secret: str) -> bool:
parts = dict(p.split("=", 1) for p in headers["NGS-Signature"].split(","))
if abs(time.time() - int(parts["t"])) > 300:
return False
expected = hmac.new(secret.encode(), f"{parts['t']}.".encode() + body, hashlib.sha256).hexdigest()
return hmac.compare_digest(expected, parts["v1"])Already recorded the narration? Send audio_url instead of a topic or a script: an https direct-download link to an MP3, M4A or WAV, up to 60 minutes and 100 MB. We write the script from your recording word for word and time the scenes to it. No voice is generated, so there's no script or voice to pay for.
{
"audio_url": "https://www.dropbox.com/scl/fi/abc123/my-voiceover.mp3?dl=1",
"title": "The history of the Panama Canal",
"style_id": "<a style from show_styles>"
}