# sensenova-u1-mcp [Health: Active]

**Category:** 🎥 Multimedia Process  
**Repository:** https://github.com/Microbiosis/sensenova-u1-mcp  
**GitHub Stars:** 2  
**npm Downloads (last month):** 3207  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/sensenova-u1-mcp

## Description
SenseNova U1 Fast MCP Server — 商汤科技文生图 API 的 MCP 服务

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "sensenova-u1-mcp": {
    "command": "npx",
    "args": ["-y","sensenova-u1-mcp"],
    "env": {
      "SENSENOVA_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `SENSENOVA_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What sensenova-u1-mcp does

The sensenova-u1-mcp MCP server exposes SenseNova U1 Fast image generation through the Model Context Protocol. It is intended for MCP-compatible clients that need to turn text descriptions into generated images. Prompts may be written in Chinese, and the README recommends detailed descriptions covering the subject, style, color, layout, elements, and any requested text.

The server provides one documented tool, `generate_image`. It accepts a required `prompt`, an optional `size`, and an optional `n` value. The default output size is `2752x1536`, and `n` defaults to one image. A single request can ask for between one and four images.

As a listing in the 🎥 Multimedia Process category, this server fits workflows that create visual assets, covers, posters, wallpapers, cards, or wide-format graphics from agent-generated prompts.

## How it works

The MCP client starts the package as a local process and supplies the SenseNova credential through its environment. When an agent invokes `generate_image`, the server sends a request to SenseNova's image-generation endpoint at `https://token.sensenova.cn/v1/images/generations`, using the `sensenova-u1-fast` model. The API response contains generated image URLs under its `data` array.

Supported sizes cover 11 aspect ratios, including 9:21, 9:16, 1:1, 4:3, 16:9, and 21:9. The available dimensions range from `1344x3136` to `3072x1376`; the complete list is defined by the tool documentation. Requests that use an unsupported size, invalid parameters, or an unavailable credential may return API errors rather than generated results.

## Setup and configuration

Create an account on the SenseNova platform and obtain an API key before starting the sensenova-u1-mcp MCP server. Set the key as `SENSENOVA_API_KEY` in the environment used by the MCP client. The package can be installed globally with `npm install -g sensenova-u1-mcp`, or launched without a global installation with `npx -y sensenova-u1-mcp`.

An MCP configuration uses `npx` as the command, passes `-y sensenova-u1-mcp` as arguments, and places the key in the server's `env` object. The README specifically notes that the command should remain `npx` and that the arguments should not contain a local absolute path. For local development, clone the repository, run `npm install`, set the same environment variable, and start it with `node index.js`.

## Tools and capabilities

`generate_image` supports:

- Detailed text prompts, including Chinese prompts
- A default `2752x1536` output size
- Eleven documented resolution choices
- One to four images per request
- Image URLs returned by the SenseNova API

The underlying API documents `prompt` and `model` as required request fields, with `sensenova-u1-fast` used as the model. `size` and `n` are optional inputs with the defaults described above.

## Limitations and notes

The README documents a quota of 1,500 requests per five-hour period and identifies HTTP 429 as the response for exceeded limits. HTTP 400 indicates invalid parameters, 403 indicates an authorization problem, and 500 indicates a server-side error. The appropriate first checks are the request values and API key; server errors may require trying again later.

The material documents image generation only. It does not describe image editing, persistent storage, post-processing, or additional MCP tools. Generated results are represented as URLs in the API response, so downstream handling of those URLs belongs to the client or agent workflow.

_Full upstream README: https://allmcps.com/mcp/sensenova-u1-mcp/readme_