seedance ai
MCP
Plug Claude, ChatGPT, Cursor and other AI apps into seedance ai. Ask for images, video or music in the chat, and our tools make it on your account.
Server URL
https://sdanceai.com/api/mcpSign in with seedance ai when your app asks, or send an sk_ key.
- Claude
ChatGPT- Cursor
- VS Code
- Claude Code
CodexGemini CLI
Claude is connected to seedance ai
Freeze this jump in bullet time, orbit the camera

Example results, made with models on seedance ai.
Connect in under a minute
Pick your app. Apps with sign-in need no key; the others use an API key you create right here.
Claude (claude.ai, Desktop and mobile)
Sign in with seedance ai, no key
- 1
Add the connector
The button opens Claude's Add custom connector box with the name and URL filled in. By hand: Customize → Connectors → + Add → Add custom connector, name it seedance ai and paste the URL.
Server URLhttps://sdanceai.com/api/mcp
- 2
Sign in to seedance ai
Click Add, then Connect. A seedance ai page opens: sign in if asked and click Allow. No key needed. It then shows in your keys list below, and deleting it there disconnects Claude.
- 3
Turn it on in a chat
In a chat click + → Connectors and switch on seedance ai. Set the read-only tools to Always allow on the connector page so Claude stops asking before checks like your balance.
Connectors added on claude.ai sync to Claude Desktop and mobile. On Team and Enterprise plans an owner may need to add it for the organization.
What your agent can do once it is plugged in
Ask in plain words. Your agent picks the tool and the model, checks the price, waits for the job and puts the seedance ai result in the chat.
Describe the shot in plain words. Your agent picks the tool, sets the options and waits for the result.
Make a 6-second slow push-in on white sneakers on a lilac set, soft shadows, 3:4.
Connect seedance ai there first.
Claude is connected to seedance ai
Make a 6-second slow push-in on white sneakers on a lilac set, soft shadows, 3:4.

One prompt in, finished media out
Ask in plain words, seedance ai runs the tools, and the result lands in your chat: images, video and music.
01 / 03
Images
from one sentence.
- 1You ask in plain words
- 2The agent picks a model
- 3The tools run
Claude

Jobs your agent can run end to end
Copy a prompt or open it in Claude. Each card shows the tool and the model it uses.
- MarketingFeaturedgenerate_video
- Commercegenerate_video
- Videogenerate_video
- Videogenerate_video
- Videogenerate_video
- Musicgenerate_music
- Marketinggenerate_video
- Videogenerate_video
- Commercegenerate_image
- Imagesgenerate_image
- Imagesgenerate_image
Every creative model, inside your chat
Image, video and music models from the app, behind one key. Your agent can call list_models for the live list.
10 tools your agent can call
Your agent sees these after you connect. Jobs run in the background, and the agent waits for the results. Full API reference →
generate_imageCreate images from a prompt, or edit and combine reference images you pass in.generate_videoMake a clip from text, a start image or a reference video.generate_musicCompose a song or an instrumental.list_modelsImage, video and audio models you can pick.get_modelEvery option a model takes: ratio, resolution, duration…estimate_costThe credit cost of a job, without running it.get_balanceYour remaining credits.wait_for_generationWait until a job finishes and return the result link.get_generationStatus and result link of one job.list_generationsYour recent jobs, newest first.
Questions
Still stuck? Ask your agent to call get_balance. If that works, you're connected.
Plug in once.
Then just ask.
Same account and credits as the app. Your agent checks the cost before it spends, and every result lands in your seedance ai creations.
https://sdanceai.com/api/mcp