Seedance MCP Integration Guide

MCP (Model Context Protocol) is a model context protocol launched by Anthropic, allowing AI models (such as Claude, GPT, etc.) to call external tools through standardized interfaces. Through the Seedance MCP Server provided by AceData Cloud, you can directly use ByteDance Seedance to generate AI videos in AI clients such as Claude Desktop, VS Code, Cursor, and more.

Feature Overview

Seedance MCP Server provides the following core features:

Text-to-video — Generate high-quality videos through text prompts
Image-to-video — Generate videos using images as references (first frame, last frame, reference image, and other modes)
Multi-model support — Supports Seedance 1.x, 1.5, 2.0, and 2.5
Multiple resolutions — Supports 480p, 720p, 1080p; some models support 4k
Multiple aspect ratios — Supports 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, and adaptive
Flexible duration — Supports 2–30 seconds or automatic duration depending on the model
Multimodal video — Supports audio generation, reference audio/video, as well as video editing and extension in 2.5
Task queries — Monitor generation progress and obtain results

Recommended: OAuth Authorization with Hosted MCP

Add the server address in clients that support remote MCP OAuth:

https://seedance.mcp.acedata.cloud/mcp

Select connect or log in, log in to Ace Data Cloud, review permissions and authorize, then enable the tools. This path does not require creating or copying an API Token first. The service supports OAuth discovery, Dynamic Client Registration (DCR), and S256 PKCE; DCR registers the client, users still need to log in and authorize, and API usage is billed as usual.

Client	Integration Method and Requirements
Claude Web or Desktop versions that provide remote connectors	Add the above address in the connector interface and authorize; availability depends on the version and plan. Do not write remote HTTP configuration into local stdio configuration files.
VS Code with native remote MCP OAuth support	Run MCP: Add Server, select HTTP, enter the address and log in as prompted; the dedicated AceDataCloud extension uses separate API Key settings.
Other clients	Confirm that the current version supports remote HTTP, OAuth/DCR, and follow that client’s configuration instructions. Configuration file formats cannot be shared across different clients.

API Token Alternative Path

Use API Token for local stdio, self-hosting, or clients that do not support remote OAuth. Log in to the Ace Data Cloud platform, select a service, and create API credentials with the required permissions. Store the Token in the client’s credential settings or local environment variables; do not put it in URLs, chat content, or repositories. HTTP Header configuration for different clients should be completed according to their version documentation. Installing open-source MCP or completing authorization does not mean that free credits are available, nor does it guarantee generation time or commercial usage rights. Review current service pricing, balance, and applicable terms before generation.

Method Two: Run Locally (pip / uvx Installation)

If you need local stdio, custom code, or self-hosting, you can start the MCP Server locally. Generation still requires an internet connection to call the API; it is not offline inference.

Installation

pip installation (recommended):

pip install mcp-seedance

Or install from source:

git clone https://github.com/AceDataCloud/SeedanceMCP.git
cd SeedanceMCP
pip install -e .

After installation, you can use the mcp-seedance command to start the service.

Use in Claude Desktop (Local)

Edit the Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the following configuration:

{
  "mcpServers": {
    "seedance": {
      "command": "mcp-seedance",
      "env": {
        "ACEDATACLOUD_API_TOKEN": "你的API Token"
      }
    }
  }
}

If using uvx (no need to install the package in advance):

{
  "mcpServers": {
    "seedance": {
      "command": "uvx",
      "args": ["mcp-seedance"],
      "env": {
        "ACEDATACLOUD_API_TOKEN": "你的API Token"
      }
    }
  }
}

Restart Claude Desktop after saving the configuration for it to take effect.

Use in VS Code (Local)

Create .vscode/mcp.json in the project root directory:

{
  "servers": {
    "seedance": {
      "command": "mcp-seedance",
      "env": {
        "ACEDATACLOUD_API_TOKEN": "你的API Token"
      }
    }
  }
}

Or use uvx:

{
  "servers": {
    "seedance": {
      "command": "uvx",
      "args": ["mcp-seedance"],
      "env": {
        "ACEDATACLOUD_API_TOKEN": "你的API Token"
      }
    }
  }
}

For other local clients, fill in command, args, and env according to the stdio configuration instructions for their current version. Do not directly treat VS Code configuration as a universal format.

Verify Results and Common Failures

Ask the client to generate a simple video work, retain the task ID and query the same task until it reaches a successful final state, then open or play the final link. A successful connection, loaded tool list, HTTP 200, task ID, and in-progress preview do not represent the final output. If you encounter 401, log in again or check the Token and permissions; if the balance is insufficient, check billing; continue querying while the task is processing; when it ultimately fails, read the error and decide whether to retry, as new generations may be billed again.

Available Tools

Tool Name	Description
`seedance_generate_video`	Generate video through text prompts
`seedance_generate_video_from_image`	Generate video using images as references
`seedance_get_task`	Query the status of a single task
`seedance_get_tasks_batch`	Query task statuses in batches
`seedance_list_models`	List all available models and their capabilities
`seedance_list_resolutions`	List available resolutions and aspect ratios
`seedance_list_actions`	List all available tools and workflow examples

Usage Examples

After configuration is complete, you can directly use natural language in the AI client to call these features, for example:

“Help me use Seedance to generate a city street time-lapse video”
“Use this photo as the first frame and generate an 8-second video”
“Generate a 1080p vertical 9:16 short video”
“Use the Seedance 1.5 Pro model to generate a video with audio”

More Information

GitHub repository: AceDataCloud/SeedanceMCP
PyPI package: mcp-seedance
API documentation: Seedance Video Generation API

​Feature Overview

​Recommended: OAuth Authorization with Hosted MCP

​API Token Alternative Path

​Method Two: Run Locally (pip / uvx Installation)

​Installation

​Use in Claude Desktop (Local)

​Use in VS Code (Local)

​Verify Results and Common Failures

​Available Tools

​Usage Examples

​More Information