Seedance 2.5 MCP Server
MCP server for generating Seedance 2.5 videos via MuAPI, enabling text-to-video, image-to-video, first/last frame, and omni reference workflows with 720p/480p resolution selection.
README
Seedance 2.5 MCP Server
A focused Model Context Protocol server for generating Seedance 2.5 Preview videos through MuAPI. It exposes the standard Seedance 2.5 workflows with explicit 720p/480p selection, rather than making an assistant search through a large general-purpose model catalog.
The implementation is informed by SamurAIGPT/muapi-mcp-server: it forwards a MuAPI key, keeps generation asynchronous, and provides both an MCP stdio transport and a small HTTP bridge.
📺 Video Tutorial
How to Access Seedance 2.5 API (Step-by-Step Guide) — a full walkthrough of getting an API key and making your first Seedance 2.5 call via MuAPI.
Related Projects
- MuAPI Seedance 2.5 — Model landing page and browser playground links for the Seedance 2.5 family.
- MuAPI MCP documentation — Hosted MCP setup and tool-use guidance.
- MuAPI access keys — Create the API key required by this server.
- Seedance-2.5-API — Python wrapper for the Seedance 2.5 API.
- Seedance-2-API — Python wrapper covering Seedance 2.0 and Seedance 2 Mini.
- seedance2.5-comfyui — Native Seedance 2.5 custom nodes and workflows for ComfyUI.
- seedance2-comfyui — Related Seedance 2 custom nodes and workflows.
- n8n-nodes-seedance2 — Seedance automation nodes for n8n workflows.
- awesome-seedance-2.5-api-prompts — Prompt and camera-control references for Seedance 2.5.
- muapi-mcp-server — Broad MuAPI MCP server reference for Claude, Cursor, and other clients.
- Open-Generative-AI — Open-source media studio that uses MuAPI for image and video workflows.
Included tools
| Tool | Purpose |
|---|---|
seedance_25_text_to_video |
Text-to-video generation at 720p or 480p |
seedance_25_image_to_video |
Animate one input image at 720p or 480p |
seedance_25_first_last_frame |
Transition between exactly two frame images |
seedance_25_omni_reference |
Combine image, video, and audio references |
muapi_predict_result |
Poll a MuAPI prediction by request_id |
muapi_account_balance |
Read the MuAPI credit balance |
Generation tools return the completed prediction when used over stdio. The HTTP bridge returns a local request ID immediately and exposes the result at /mcp/predictions/{request_id} or as server-sent events.
Quick start
git clone https://github.com/Anil-matcha/seedance-2.5-mcp.git
cd seedance-2.5-mcp
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
export MUAPI_API_KEY=your_muapi_key_here
Run the standards-based stdio server for an MCP client:
python mcp_stdio.py
Run the local HTTP bridge instead:
python mcp_server.py
# http://localhost:8000/docs
MUAPIAPP_API_KEY is accepted as a backwards-compatible alternative. MUAPI_BASE_URL, MUAPI_HTTP_TIMEOUT, MUAPI_POLL_INTERVAL, and MUAPI_POLL_TIMEOUT can be used for local testing or a compatible MuAPI deployment.
Claude Code
With MUAPI_API_KEY exported in the shell used by Claude Code:
claude mcp add seedance-2.5 -- python /absolute/path/to/seedance-2.5-mcp/mcp_stdio.py
Cursor, Windsurf, and Claude Desktop
Copy the mcp.json entry and replace the placeholder path. Keep the API key in the client environment; do not commit it.
{
"mcpServers": {
"seedance-2.5": {
"command": "python",
"args": ["/absolute/path/to/seedance-2.5-mcp/mcp_stdio.py"],
"env": {
"MUAPI_API_KEY": "${MUAPI_API_KEY}"
}
}
}
}
HTTP bridge
The HTTP process is useful for local integrations that need a simple request/poll surface:
curl http://localhost:8000/mcp/tools
curl -X POST http://localhost:8000/mcp/tools/seedance_25_text_to_video/call \
-H "Authorization: Bearer $MUAPI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"arguments": {
"prompt": "A red paper boat drifting through a sunlit forest stream, cinematic camera movement",
"duration": 5,
"aspect_ratio": "16:9",
"resolution": "720p"
}
}'
Poll the returned local ID:
curl http://localhost:8000/mcp/predictions/REQUEST_ID
curl -N http://localhost:8000/mcp/predictions/REQUEST_ID/stream
The bridge allowlists all upstream paths and never accepts an arbitrary endpoint URL. Generation and polling consume MuAPI credits according to the selected resolution and workflow.
Model notes
Seedance 2.5 Preview supports clips from 4 to 30 seconds. The standard routes support 720p and the dedicated 480p routes are intended for faster, lower-cost previews. The first/last-frame tool requires exactly two images. Omni reference accepts up to 20 images, 6 video clips, and 6 audio files.
Docker
docker build -t seedance-2-5-mcp .
docker run --rm -p 8000:8000 -e MUAPI_API_KEY="$MUAPI_API_KEY" seedance-2-5-mcp
Development
python -m unittest discover -s tests -v
python -m compileall -q server_core.py mcp_server.py mcp_stdio.py
License
MIT. See LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
