Wellknown found it in public sources; nobody has proven control of it yet. Claiming takes one click if the repository is under your GitHub account, or a small file on your domain otherwise. Verified owners get the badge, 15-minute checks, status alerts, edits that outrank crawled data, and a ranking boost.
Agents can do it too: POST https://wellknown.network/api/v1/claims with {"agent":"mcp-server-bailian-image","method":"well_known_file"} — machine-readable steps at claim.json, guide at /docs/claim.
Everything here was measured by our prober or read from a registry. Nothing is self-reported.
Attributed to the source that supplied each field. Treated as claims, not facts.
# MCP Server Bailian Image 阿里云百炼-通义万相图像生成MCP服务器 ## 功能特性 - 🎨 **文生图功能**: 基于通义万相2.1/2.2文生图V2版API - 🚀 **多模型支持**: 支持wan2.2-t2i-flash、wan2.2-t2i-plus等多种模型 - 🎯 **精确控制**: 支持正向提示词、反向提示词、图像尺寸、生成数量等参数 - ⚡ **同步调用**: 使用DashScope SDK进行同步调用,简化使用流程 - 🔧 **易于集成**: 标准MCP协议,可与支持MCP的AI助手无缝集成 ## 安装 ### 从PyPI安装 ```bash pip install mcp-server-bailian-image ``` ### 从源码安装 ```bash git clone https://github.com/aliyun/mcp-server-bailian-image.git cd mcp-server-bailian-image pip install -e . ``` ## 配置 ### 1. 获取API Key 1. 访问[阿里云百炼控制台](https://bailian.console.aliyun.com/) 2. 创建应用并获取API Key 3. 确保已开通通义万相文生图服务 ### 2. 设置环境变量 ```bash # Windows set DASHSCOPE_API_KEY=your_api_key_here # Linux/macOS export DASHSCOPE_API_KEY=your_api_key_here ``` ## 使用方法 ### 启动MCP服务器 ```bash # 使用uvx启动(推荐) uvx mcp-server-bailian-image # 或者直接传入API Key uvx mcp-server-bailian-image --api-key your_api_key_here ``` ### 在Claude Desktop中配置 在Claude Desktop的配置文件中添加以下配置: ```json { "mcpServers": { "bailian-image": { "command": "uvx", "args": ["mcp-server-bailian-image"], "env": { "DASHSCOPE_API_KEY": "your_api_key_here" } } } } ``` ## 支持的工具 ### text2imagev2 基于通义万相文生图V2版API的文生图工具。 **参数说明:** - `prompt` (必填): 正向提示词,描述期望生成的图像内容 - `negative_prompt` (可选): 反向提示词,描述不希望出现的内容 - `model` (可选): 模型名称,默认为 `wan2.2-t2i-flash` - `size` (可选): 图像尺寸,默认为 `1024*1024` - `n` (可选): 生成图片数量,取值范围1-4,默认为1 **支持的模型:** - `wan2.2-t2i-flash`: 万相2.2极速版(推荐) - `wan2.2-t2i-plus`: 万相2.2专业版(推荐) - `wanx2.1-t2i-turbo`: 万相2.1极速版 - `wanx2.1-t2i-plus`: 万相2.1专业版 - `wanx2.0-t2i-turbo`: 万相2.0极速版 **支持的图像尺寸:** - `1024*1024` (默认) - `720*1280` - `1280*720` - `1024*576` - `576*1024` **使用示例:** ```python # 基础文生图 result = await call_tool("text2imagev2", { "prompt": "一只坐着的橘黄色的猫,表情愉悦,活泼可爱,逼真准确" }) # 使用反向提示词 result = await call_tool("text2imagev2", { "prompt": "雪地,白色小教堂,极光,冬日场景,柔和的光线", "negative_prompt": "人物", "model": "wan2.2-t2i-plus", "size": "1280*720", "n": 2 }) ``` ## 错误处理 常见错误及解决方…
Mapped onto the structured taxonomy from declared text and observed tool names. Confidence shown for derived entries.
Every source is kept verbatim. Field changes are logged as events.