Your agent just
got a film studio.
The Supervivid MCP server puts every frontier video, image, and music model one tool call away from Claude, Cursor, ChatGPT — or any agent you build.
>Make a 10-second hero video of a surfer inside a glassy wave at golden hour, then score it with warm analog synths.
⏺supervivid · generate_video
model: seedance-2.5duration: 10sratio: 16:9resolution: 1080p
wave-hero.mp4· 10s · 1080p · saved to ./public/videos
⏺supervivid · generate_music"warm analog synth swell, 10s"
wave-hero-score.mp3· mixed under the master
Done — hero video is live in your project.
Connect in 60 seconds
One config block, one API key. The server speaks standard MCP over HTTP, so there's nothing to install and nothing to babysit.
claude mcp add --transport http supervivid https://mcp.supervivid.ai/mcp \
--header "Authorization: Bearer $SUPERVIVID_API_KEY"The whole suite, as tools
Everything Supervivid can do in the browser, your agent can do over MCP — same models, same credits, same quality.
generate_video
Text or image to cinematic video. Your agent picks the model, duration, ratio, and resolution.
generate_image
Stills in any style — photoreal, anime, 3D render — up to 4K, single frames or batches.
generate_music
Original tracks and scores from a text brief, from 15-second stings to full songs.
face_swap
Swap any face into existing footage with one reference image, seamlessly tracked.
lip_sync
Match any speaker's lips to new dialogue or a translated voice track.
upscale
Enhance any render or upload to native 4K with detail-preserving upscaling.
list_models
Discover every available model with its capabilities, so agents always pick the right one.
get_generation
Poll render status and fetch finished files as URLs or straight into the workspace.
From prompt to premiere
Connect
Paste one config block into your MCP client and drop in your Supervivid API key. No SDK, no wrapper code.
Direct
Describe the shot in plain English. Your agent chooses the model and settings — or you specify them, down to the frame.
Ship
Finished video, images, and music come back as files and URLs, ready to drop into your site, feed, or edit.
MCP questions, answered
The short version of how the Supervivid MCP server works. Something else on your mind?
Talk to supportWhat is MCP?
The Model Context Protocol is an open standard that lets AI agents call external tools. The Supervivid MCP server exposes our entire creative suite — video, image, music, face swap, lip sync, upscaling — as tools any MCP-compatible agent can use.
Which agents and apps can connect?
Anything that speaks MCP over HTTP: Claude Code, Claude Desktop, Cursor, Windsurf, ChatGPT, VS Code, Zed, Gemini CLI, and custom agents built on the official MCP SDKs. If your client supports remote MCP servers, it works with Supervivid.
How is usage billed?
Generations through MCP draw from the same credit pool as your Supervivid plan — one subscription covers the web app and the MCP server. Your daily free credits work here too.
Can my agent choose which model to use?
Yes. Agents can call list_models to see every available model with its strengths and pricing, then pick the best fit — or you can pin a specific model like Seedance 2.5 or Flux 3 in your prompt.
Where do generated files end up?
Every tool returns a hosted URL immediately, and coding agents like Claude Code and Cursor can download results straight into your project — so a hero video can land in ./public/videos without you leaving the chat.
How do I get an API key?
Create a free Supervivid account and generate a key from your account settings. Keys are scoped to your workspace and can be revoked or rotated at any time.
Give your agent
superpowers.
Grab a key, paste one config block, and your agent is rendering cinematic video before your coffee cools.
Free credits every day