Doubao Image/Video Generation MCP Server

Instructions

Implementation Reference

• src/tools/generateImage.ts:41-110 (handler)

  The main handler function that executes the image generation logic by constructing the request to the Doubao API and handling the response.
  export async function generateImage( apiKey: string, options: GenerateImageOptions ): Promise<ImageGenerationResponse> { const { prompt, model, endpoint_id, size = "1920x2160", image_url, ref_image_urls, req_key, watermark = false, // 默认不加水印 } = options; // 优先使用 endpoint_id，如果没有则使用 model const modelOrEndpoint = endpoint_id || model || "doubao-seedream-4-5"; // 构建请求体 const requestBody: Record<string, any> = { model: modelOrEndpoint, prompt, size, watermark, // 添加水印参数 }; // 添加可选参数 if (image_url) { requestBody.image_url = image_url; } if (ref_image_urls && ref_image_urls.length > 0) { requestBody.ref_image_urls = ref_image_urls; } if (req_key) { requestBody.req_key = req_key; } try { const response = await fetch(`${BASE_URL}/images/generations`, { method: "POST", headers: { "Content-Type": "application/json", Authorization: `Bearer ${apiKey}`, }, body: JSON.stringify(requestBody), }); const result: ImageGenerationResponse = await response.json(); // 检查是否有错误 if (result.error) { throw new Error( `图片生成失败: ${result.error.message || "未知错误"}` ); } // 检查 HTTP 状态码 if (!response.ok) { throw new Error(`HTTP 错误: ${response.statusText}`); } return result; } catch (error) { throw new Error( `图片生成请求失败: ${error instanceof Error ? error.message : String(error)}` ); } }
• src/tools/generateImage.ts:6-15 (schema)

  TypeScript interface defining the input parameters for the generateImage handler.
  interface GenerateImageOptions { prompt: string; model?: string; size?: string; image_url?: string; ref_image_urls?: string[]; req_key?: string; endpoint_id?: string; // 推理接入点 ID watermark?: boolean; // 是否添加水印，默认 false }
• src/index.ts:78-122 (schema)

  MCP protocol inputSchema defining the parameters and validation for the generate_image tool.
  inputSchema: { type: "object", properties: { prompt: { type: "string", description: "图片描述文本", }, endpoint_id: { type: "string", description: "推理接入点 ID (在火山引擎控制台创建)", }, model: { type: "string", description: `模型选择，默认: ${DEFAULT_IMAGE_MODEL}`, enum: ["doubao-seedream-4-5", "doubao-seedream-3-0-t2i"], }, size: { type: "string", description: "图片尺寸，默认: 1920x2160 (注意: 豆包 API 要求图片至少 3686400 像素)", enum: [ "1920x2160", "1920x2560", "2160x3840", ], }, image_url: { type: "string", description: "参考图片 URL (图生图)", }, ref_image_urls: { type: "array", items: { type: "string" }, description: "多张参考图片 URL 数组 (多图融合)", }, req_key: { type: "string", description: "请求标识", }, watermark: { type: "boolean", description: "是否添加水印，默认: false", }, }, required: ["prompt"], },
• src/index.ts:54-123 (registration)

  Registration of the generate_image tool in the ListToolsRequestHandler, including name, description, and schema.
  name: "generate_image", description: `使用豆包 Seedream 模型生成图片 支持功能: - 文生图: 使用文本提示词生成图片 - 图生图: 使用输入图片和提示词生成新图片 - 多图融合: 使用多张参考图片融合生成新图片 参数说明: - prompt: 图片描述文本 (必需) - endpoint_id: 推理接入点 ID (推荐) 在火山引擎控制台创建推理接入点后获取的 Endpoint ID - model: 模型名称 (可选，默认: ${DEFAULT_IMAGE_MODEL}) * doubao-seedream-4-5: 最新 4.0 模型，支持 4K 分辨率 * doubao-seedream-3-0-t2i: 3.0 文生图模型 注意: 直接使用模型名称可能需要账户权限，推荐使用 endpoint_id - size: 图片尺寸 (可选，默认: 1024x1024) 支持的尺寸: 1024x1024, 1024x1792, 1024x768, 768x1024, 1792x1024, 512x512, 512x768, 768x512 - image_url: 参考图片 URL (可选，用于图生图) - ref_image_urls: 多张参考图片 URL 数组 (可选，用于多图融合) - req_key: 请求标识 (可选，用于追踪) 重要提示: 如果遇到 InvalidEndpointOrModel.NotFound 错误，请在火山引擎控制台创建推理接入点，并使用 endpoint_id 参数`, inputSchema: { type: "object", properties: { prompt: { type: "string", description: "图片描述文本", }, endpoint_id: { type: "string", description: "推理接入点 ID (在火山引擎控制台创建)", }, model: { type: "string", description: `模型选择，默认: ${DEFAULT_IMAGE_MODEL}`, enum: ["doubao-seedream-4-5", "doubao-seedream-3-0-t2i"], }, size: { type: "string", description: "图片尺寸，默认: 1920x2160 (注意: 豆包 API 要求图片至少 3686400 像素)", enum: [ "1920x2160", "1920x2560", "2160x3840", ], }, image_url: { type: "string", description: "参考图片 URL (图生图)", }, ref_image_urls: { type: "array", items: { type: "string" }, description: "多张参考图片 URL 数组 (多图融合)", }, req_key: { type: "string", description: "请求标识", }, watermark: { type: "boolean", description: "是否添加水印，默认: false", }, }, required: ["prompt"], }, },
• src/index.ts:240-255 (registration)

  Dispatch handler in CallToolRequestHandler that invokes the generateImage function for the generate_image tool.
  case "generate_image": { // 如果没有提供 endpoint_id 且环境变量中有，则使用环境变量的值 const imageArgs = { ...args }; if (!imageArgs.endpoint_id && DEFAULT_IMAGE_ENDPOINT_ID) { imageArgs.endpoint_id = DEFAULT_IMAGE_ENDPOINT_ID; } const result = await generateImage(API_KEY, imageArgs as any); return { content: [ { type: "text", text: JSON.stringify(result, null, 2), }, ], }; }