@bashcat/ai-image-chat-mcp 2.3.5 → 3.0.0

package/dist/index.js

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
-import { Server } from "@modelcontextprotocol/sdk/server/index.js";
+import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
 import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
-import { CallToolRequestSchema, ErrorCode, ListToolsRequestSchema, McpError, } from "@modelcontextprotocol/sdk/types.js";
+import { z } from "zod";
 import dotenv from "dotenv";
 import axios from "axios";
 import * as fs from "fs";
@@ -13,990 +13,422 @@ const require = createRequire(import.meta.url);
 const packageJson = require("../package.json");
 // 載入環境變數
 dotenv.config();
-class AIImageChatMCPServer {
-    server;
-    apiKey;
-    baseUrl;
-    saveDirectory;
-    dashScopeApiKey;
-    constructor() {
-        this.server = new Server({
-            name: "ai-image-chat-mcp-server",
-            version: packageJson.version,
-        }, {
-            capabilities: {
-                tools: {},
-            },
-        });
-        // 從環境變數獲取 API 配置
-        this.apiKey = process.env.AI_API_KEY || "";
-        this.baseUrl = process.env.AI_API_BASE_URL || "https://api.laozhang.ai/v1";
-        this.saveDirectory = process.env.AI_IMAGE_SAVE_PATH || path.join(os.homedir(), "generated_images");
-        // 支援 ALI_API_KEY 和 DASHSCOPE_API_KEY 兩種環境變數名稱
-        this.dashScopeApiKey = process.env.ALI_API_KEY || process.env.DASHSCOPE_API_KEY || "";
-        if (!this.apiKey) {
-            throw new Error("AI_API_KEY 環境變數未設定");
-        }
-        // 確保保存目錄存在
-        if (!fs.existsSync(this.saveDirectory)) {
-            fs.mkdirSync(this.saveDirectory, { recursive: true });
-        }
-        this.setupToolHandlers();
-    }
-    generateFilename(prompt, outputFormat = 'jpg') {
-        // 清理 prompt 作為檔名，移除特殊字符並限制長度
-        const cleanPrompt = prompt
-            .replace(/[^\w\s\u4e00-\u9fff]/g, '') // 保留字母、數字、空格和中文字符
-            .replace(/\s+/g, '-') // 將空格替換為連字符
-            .substring(0, 50); // 限制長度為50字符
-        const timestamp = new Date().toISOString().replace(/[:.]/g, '-');
-        // 確保輸出格式為小寫
-        const extension = outputFormat.toLowerCase() === 'jpeg' ? 'jpg' : outputFormat.toLowerCase();
-        return `${cleanPrompt}-${timestamp}.${extension}`;
+// ===== 配置 =====
+const apiKey = process.env.AI_API_KEY || "";
+const baseUrl = process.env.AI_API_BASE_URL || "https://api.laozhang.ai/v1";
+const saveDirectory = process.env.AI_IMAGE_SAVE_PATH || path.join(os.homedir(), "generated_images");
+const dashScopeApiKey = process.env.ALI_API_KEY || process.env.DASHSCOPE_API_KEY || "";
+if (!apiKey) {
+    throw new Error("AI_API_KEY 環境變數未設定");
+}
+if (!fs.existsSync(saveDirectory)) {
+    fs.mkdirSync(saveDirectory, { recursive: true });
+}
+// ===== 工具函式 =====
+function generateFilename(prompt, outputFormat = "jpg") {
+    const cleanPrompt = prompt
+        .replace(/[^\w\s\u4e00-\u9fff]/g, "")
+        .replace(/\s+/g, "-")
+        .substring(0, 50);
+    const timestamp = new Date().toISOString().replace(/[:.]/g, "-");
+    const extension = outputFormat.toLowerCase() === "jpeg" ? "jpg" : outputFormat.toLowerCase();
+    return `${cleanPrompt}-${timestamp}.${extension}`;
+}
+function formatPathForDisplay(filePath) {
+    const homeDir = os.homedir();
+    return filePath.startsWith(homeDir) ? filePath.replace(homeDir, "~") : filePath;
+}
+async function downloadAndSaveImage(imageUrl, filename, outputFormat = "jpg") {
+    const response = await axios.get(imageUrl, { responseType: "arraybuffer" });
+    const filePath = path.join(saveDirectory, filename);
+    const fmt = outputFormat.toLowerCase() === "jpeg" ? "jpg" : outputFormat.toLowerCase();
+    const sharpInstance = sharp(Buffer.from(response.data));
+    if (fmt === "png") {
+        await sharpInstance.png().toFile(filePath);
     }
-    formatSaveDirectoryForDisplay() {
-        // 如果是預設的 home 目錄路徑，顯示更友好的格式
-        const homeDir = os.homedir();
-        if (this.saveDirectory.startsWith(homeDir)) {
-            return this.saveDirectory.replace(homeDir, '~');
-        }
-        return this.saveDirectory;
+    else if (fmt === "webp") {
+        await sharpInstance.webp({ quality: 90 }).toFile(filePath);
     }
-    formatFilePathForDisplay(filePath) {
-        // 將絕對路徑轉換為相對於 home 目錄的顯示格式
-        const homeDir = os.homedir();
-        if (filePath.startsWith(homeDir)) {
-            return filePath.replace(homeDir, '~');
-        }
-        return filePath;
+    else {
+        await sharpInstance.jpeg({ quality: 90 }).toFile(filePath);
     }
-    async downloadAndSaveImage(imageUrl, filename, outputFormat = 'jpg') {
-        try {
-            const response = await axios.get(imageUrl, { responseType: 'arraybuffer' });
-            const filePath = path.join(this.saveDirectory, filename);
-            // 根據輸出格式處理圖片
-            const normalizedFormat = outputFormat.toLowerCase() === 'jpeg' ? 'jpg' : outputFormat.toLowerCase();
-            if (normalizedFormat === 'jpg' || normalizedFormat === 'jpeg') {
-                // 轉換為 JPG 格式
-                await sharp(Buffer.from(response.data))
-                    .jpeg({ quality: 90 })
-                    .toFile(filePath);
-            }
-            else if (normalizedFormat === 'png') {
-                // 保持 PNG 格式
-                await sharp(Buffer.from(response.data))
-                    .png()
-                    .toFile(filePath);
-            }
-            else if (normalizedFormat === 'webp') {
-                // 轉換為 WebP 格式
-                await sharp(Buffer.from(response.data))
-                    .webp({ quality: 90 })
-                    .toFile(filePath);
-            }
-            else {
-                // 預設轉換為 JPG
-                await sharp(Buffer.from(response.data))
-                    .jpeg({ quality: 90 })
-                    .toFile(filePath);
-            }
-            return filePath;
-        }
-        catch (error) {
-            throw new Error(`圖片下載或轉換失敗: ${error instanceof Error ? error.message : String(error)}`);
-        }
-    }
-    async saveBase64Image(base64Data, filename, outputFormat = 'jpg') {
-        try {
-            // 解析 data URI 格式：data:image/png;base64,iVBORw0KGgoAAAANSU...
-            const matches = base64Data.match(/^data:image\/([a-zA-Z]+);base64,(.+)$/);
-            if (!matches) {
-                throw new Error('無效的 base64 圖片格式');
-            }
-            const [, , base64String] = matches;
-            const buffer = Buffer.from(base64String, 'base64');
-            const filePath = path.join(this.saveDirectory, filename);
-            // 根據輸出格式處理圖片
-            const normalizedFormat = outputFormat.toLowerCase() === 'jpeg' ? 'jpg' : outputFormat.toLowerCase();
-            if (normalizedFormat === 'jpg' || normalizedFormat === 'jpeg') {
-                // 轉換為 JPG 格式
-                await sharp(buffer)
-                    .jpeg({ quality: 90 })
-                    .toFile(filePath);
-            }
-            else if (normalizedFormat === 'png') {
-                // 保持 PNG 格式
-                await sharp(buffer)
-                    .png()
-                    .toFile(filePath);
-            }
-            else if (normalizedFormat === 'webp') {
-                // 轉換為 WebP 格式
-                await sharp(buffer)
-                    .webp({ quality: 90 })
-                    .toFile(filePath);
-            }
-            else {
-                // 預設轉換為 JPG
-                await sharp(buffer)
-                    .jpeg({ quality: 90 })
-                    .toFile(filePath);
-            }
-            return filePath;
-        }
-        catch (error) {
-            throw new Error(`Base64 圖片保存失敗: ${error instanceof Error ? error.message : String(error)}`);
-        }
+    return filePath;
+}
+async function saveBase64Image(base64Data, filename, outputFormat = "jpg") {
+    const matches = base64Data.match(/^data:image\/([a-zA-Z]+);base64,(.+)$/);
+    if (!matches) {
+        throw new Error("無效的 base64 圖片格式");
     }
-    extractImageUrlFromContent(content) {
-        // 嘗試從回應中提取圖片 URL
-        const urlPatterns = [
-            /https?:\/\/[^\s\)]+\.(?:jpg|jpeg|png|gif|webp)/gi,
-            /!\[.*?\]\((https?:\/\/[^\)]+)\)/gi,
-            /https?:\/\/[^\s\)]+/gi
-        ];
-        for (const pattern of urlPatterns) {
-            const matches = content.match(pattern);
-            if (matches && matches.length > 0) {
-                const url = matches[0].replace(/^\!\[.*?\]\(/, '').replace(/\)$/, '');
-                if (url.match(/\.(jpg|jpeg|png|gif|webp)(\?.*)?$/i)) {
-                    return url;
-                }
-            }
-        }
-        return null;
+    const buffer = Buffer.from(matches[2], "base64");
+    const filePath = path.join(saveDirectory, filename);
+    const fmt = outputFormat.toLowerCase() === "jpeg" ? "jpg" : outputFormat.toLowerCase();
+    const sharpInstance = sharp(buffer);
+    if (fmt === "png") {
+        await sharpInstance.png().toFile(filePath);
     }
-    extractMultipleImageUrls(content) {
-        // 提取所有圖片 URL
-        const urlPatterns = [
-            /https?:\/\/[^\s\)]+\.(?:jpg|jpeg|png|gif|webp)/gi,
-            /!\[.*?\]\((https?:\/\/[^\)]+)\)/gi,
-            /https?:\/\/[^\s\)]+/gi
-        ];
-        const urls = [];
-        for (const pattern of urlPatterns) {
-            const matches = content.match(pattern);
-            if (matches) {
-                for (const match of matches) {
-                    const url = match.replace(/^\!\[.*?\]\(/, '').replace(/\)$/, '');
-                    if (url.match(/\.(jpg|jpeg|png|gif|webp)(\?.*)?$/i)) {
-                        urls.push(url);
-                    }
-                }
-            }
-        }
-        // 去重並返回
-        return [...new Set(urls)];
+    else if (fmt === "webp") {
+        await sharpInstance.webp({ quality: 90 }).toFile(filePath);
     }
-    extractBase64Images(content) {
-        // 提取所有 base64 圖片
-        const base64Pattern = /data:image\/[a-zA-Z]+;base64,[A-Za-z0-9+/=]+/gi;
-        const matches = content.match(base64Pattern);
-        return matches || [];
+    else {
+        await sharpInstance.jpeg({ quality: 90 }).toFile(filePath);
     }
-    async processImageContent(content, prompt, outputFormat = 'jpg') {
-        const imageUrls = this.extractMultipleImageUrls(content);
-        const base64Images = this.extractBase64Images(content);
-        const savedImages = [];
-        // 處理 base64 圖片
-        if (base64Images.length > 0) {
-            for (let i = 0; i < base64Images.length; i++) {
-                try {
-                    const filename = this.generateFilename(`${prompt}-${i + 1}`, outputFormat);
-                    const savedPath = await this.saveBase64Image(base64Images[i], filename, outputFormat);
-                    savedImages.push(savedPath);
-                }
-                catch (error) {
-                    console.error(`保存第 ${i + 1} 張 base64 圖片失敗:`, error);
+    return filePath;
+}
+function extractMultipleImageUrls(content) {
+    const urlPatterns = [
+        /https?:\/\/[^\s\)]+\.(?:jpg|jpeg|png|gif|webp)/gi,
+        /!\[.*?\]\((https?:\/\/[^\)]+)\)/gi,
+        /https?:\/\/[^\s\)]+/gi,
+    ];
+    const urls = [];
+    for (const pattern of urlPatterns) {
+        const matches = content.match(pattern);
+        if (matches) {
+            for (const match of matches) {
+                const url = match.replace(/^\!\[.*?\]\(/, "").replace(/\)$/, "");
+                if (url.match(/\.(jpg|jpeg|png|gif|webp)(\?.*)?$/i)) {
+                    urls.push(url);
                 }
             }
         }
-        return {
-            imageUrls,
-            savedImages,
-            hasBase64: base64Images.length > 0
-        };
-    }
-    // 阿里雲 DashScope 創建任務
-    async createDashScopeTask(prompt, negativePrompt, options) {
-        if (!this.dashScopeApiKey) {
-            throw new Error("ALI_API_KEY 或 DASHSCOPE_API_KEY 環境變數未設定");
-        }
-        const requestData = {
-            model: options?.model || "wanx2.1-t2i-turbo",
-            input: {
-                prompt,
-                negative_prompt: negativePrompt || "人物",
-            },
-            parameters: {
-                size: options?.size || "1024*1024",
-                n: options?.n || 1,
-                seed: options?.seed,
-                prompt_extend: options?.promptExtend !== false,
-                watermark: options?.watermark || false,
-            },
-        };
-        try {
-            const response = await axios.post("https://dashscope.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis", requestData, {
-                headers: {
-                    "Content-Type": "application/json",
-                    "Authorization": `Bearer ${this.dashScopeApiKey}`,
-                    "X-DashScope-Async": "enable",
-                },
-            });
-            return response.data;
-        }
-        catch (error) {
-            if (axios.isAxiosError(error)) {
-                const axiosError = error;
-                const errorMessage = axiosError.response?.data?.message || axiosError.message;
-                throw new Error(`阿里雲 DashScope 創建任務失敗: ${errorMessage}`);
-            }
-            throw new Error(`創建任務時發生未知錯誤: ${error}`);
-        }
     }
-    // 阿里雲 DashScope 查詢任務結果
-    async queryDashScopeTask(taskId) {
-        if (!this.dashScopeApiKey) {
-            throw new Error("ALI_API_KEY 或 DASHSCOPE_API_KEY 環境變數未設定");
-        }
+    return [...new Set(urls)];
+}
+function extractBase64Images(content) {
+    const matches = content.match(/data:image\/[a-zA-Z]+;base64,[A-Za-z0-9+/=]+/gi);
+    return matches || [];
+}
+async function processImageContent(content, prompt, outputFormat = "jpg") {
+    const imageUrls = extractMultipleImageUrls(content);
+    const base64Images = extractBase64Images(content);
+    const savedImages = [];
+    for (let i = 0; i < base64Images.length; i++) {
         try {
-            const response = await axios.get(`https://dashscope.aliyuncs.com/api/v1/tasks/${taskId}`, {
-                headers: {
-                    "Authorization": `Bearer ${this.dashScopeApiKey}`,
-                },
-            });
-            return response.data;
+            const filename = generateFilename(`${prompt}-${i + 1}`, outputFormat);
+            const savedPath = await saveBase64Image(base64Images[i], filename, outputFormat);
+            savedImages.push(savedPath);
         }
         catch (error) {
-            if (axios.isAxiosError(error)) {
-                const axiosError = error;
-                const errorMessage = axiosError.response?.data?.message || axiosError.message;
-                throw new Error(`阿里雲 DashScope 查詢任務失敗: ${errorMessage}`);
-            }
-            throw new Error(`查詢任務時發生未知錯誤: ${error}`);
+            console.error(`保存第 ${i + 1} 張 base64 圖片失敗:`, error);
         }
     }
-    // 阿里雲 DashScope 完整生圖流程
-    async generateImageWithDashScope(prompt, negativePrompt, options) {
-        // 步驟1: 創建任務
-        const createResponse = await this.createDashScopeTask(prompt, negativePrompt, options);
-        const taskId = createResponse.output.task_id;
-        if (!taskId) {
-            throw new Error("創建任務失敗，未獲取到任務ID");
-        }
-        // 步驟2: 輪詢查詢結果
-        const maxWaitMinutes = options?.maxWaitMinutes || 5; // 預設最多等待5分鐘
-        const maxAttempts = maxWaitMinutes * 6; // 每10秒查詢一次
-        let attempts = 0;
-        while (attempts < maxAttempts) {
-            const queryResponse = await this.queryDashScopeTask(taskId);
-            const taskStatus = queryResponse.output.task_status;
-            if (taskStatus === "SUCCEEDED") {
-                const results = queryResponse.output.results || [];
-                return results;
-            }
-            else if (taskStatus === "FAILED") {
-                throw new Error(`圖片生成失敗: ${queryResponse.output.message || "未知錯誤"}`);
-            }
-            else if (taskStatus === "CANCELED") {
-                throw new Error("任務已被取消");
-            }
-            // 等待10秒後再次查詢
-            await new Promise(resolve => setTimeout(resolve, 10000));
-            attempts++;
-        }
-        throw new Error(`圖片生成超時，等待時間超過 ${maxWaitMinutes} 分鐘`);
+    return { imageUrls, savedImages, hasBase64: base64Images.length > 0 };
+}
+// ===== DashScope API =====
+async function createDashScopeTask(prompt, negativePrompt, options) {
+    if (!dashScopeApiKey) {
+        throw new Error("ALI_API_KEY 或 DASHSCOPE_API_KEY 環境變數未設定");
     }
-    // 阿里雲 DashScope 視頻生成創建任務
-    async createVideoGenerationTask(imgUrl, prompt, options) {
-        if (!this.dashScopeApiKey) {
-            throw new Error("ALI_API_KEY 或 DASHSCOPE_API_KEY 環境變數未設定");
-        }
-        const requestData = {
-            model: options?.model || "wanx2.1-i2v-turbo",
-            input: {
-                img_url: imgUrl,
-                prompt: prompt,
-                template: options?.template,
+    const requestData = {
+        model: options?.model || "wanx2.1-t2i-turbo",
+        input: {
+            prompt,
+            negative_prompt: negativePrompt || "人物",
+        },
+        parameters: {
+            size: options?.size || "1024*1024",
+            n: options?.n || 1,
+            seed: options?.seed,
+            prompt_extend: options?.promptExtend !== false,
+            watermark: options?.watermark || false,
+        },
+    };
+    try {
+        const response = await axios.post("https://dashscope.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis", requestData, {
+            headers: {
+                "Content-Type": "application/json",
+                Authorization: `Bearer ${dashScopeApiKey}`,
+                "X-DashScope-Async": "enable",
             },
-            parameters: {
-                resolution: options?.resolution || "720P",
-                duration: options?.duration || 5,
-                prompt_extend: options?.promptExtend !== false,
-                seed: options?.seed,
-            },
-        };
-        try {
-            const response = await axios.post("https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis", requestData, {
-                headers: {
-                    "Content-Type": "application/json",
-                    "Authorization": `Bearer ${this.dashScopeApiKey}`,
-                    "X-DashScope-Async": "enable",
-                },
-            });
-            return response.data;
-        }
-        catch (error) {
-            if (axios.isAxiosError(error)) {
-                const axiosError = error;
-                const errorMessage = axiosError.response?.data?.message || axiosError.message;
-                throw new Error(`阿里雲 DashScope 視頻生成任務創建失敗: ${errorMessage}`);
-            }
-            throw new Error(`創建視頻生成任務時發生未知錯誤: ${error}`);
-        }
+        });
+        return response.data;
     }
-    // 阿里雲 DashScope 視頻生成完整流程
-    async generateVideoWithDashScope(imgUrl, prompt, options) {
-        // 步驟1: 創建任務
-        const createResponse = await this.createVideoGenerationTask(imgUrl, prompt, options);
-        const taskId = createResponse.output.task_id;
-        if (!taskId) {
-            throw new Error("創建視頻生成任務失敗，未獲取到任務ID");
+    catch (error) {
+        if (axios.isAxiosError(error)) {
+            const errorMessage = error.response?.data?.message || error.message;
+            throw new Error(`阿里雲 DashScope 創建任務失敗: ${errorMessage}`);
         }
-        // 步驟2: 輪詢查詢結果
-        const maxWaitMinutes = options?.maxWaitMinutes || 15; // 預設最多等待15分鐘（視頻生成較慢）
-        const maxAttempts = maxWaitMinutes * 6; // 每10秒查詢一次
-        let attempts = 0;
-        while (attempts < maxAttempts) {
-            const queryResponse = await this.queryDashScopeTask(taskId);
-            const taskStatus = queryResponse.output.task_status;
-            if (taskStatus === "SUCCEEDED") {
-                const videoUrl = queryResponse.output.video_url;
-                if (!videoUrl) {
-                    throw new Error("任務完成但未獲取到視頻URL");
-                }
-                return videoUrl;
-            }
-            else if (taskStatus === "FAILED") {
-                throw new Error(`視頻生成失敗: ${queryResponse.output.message || "未知錯誤"}`);
-            }
-            else if (taskStatus === "CANCELED") {
-                throw new Error("視頻生成任務已被取消");
-            }
-            // 等待10秒後再次查詢
-            await new Promise(resolve => setTimeout(resolve, 10000));
-            attempts++;
-        }
-        throw new Error(`視頻生成超時，等待時間超過 ${maxWaitMinutes} 分鐘`);
+        throw new Error(`創建任務時發生未知錯誤: ${error}`);
     }
-    setupToolHandlers() {
-        // 列出可用工具
-        this.server.setRequestHandler(ListToolsRequestSchema, async () => {
-            return {
-                tools: [
-                    {
-                        name: "generate_image",
-                        description: "統一圖像生成工具，支援通義萬相和老張API多種模型",
-                        inputSchema: {
-                            type: "object",
-                            properties: {
-                                prompt: {
-                                    type: "string",
-                                    description: "圖像生成的提示詞",
-                                },
-                                model: {
-                                    type: "string",
-                                    description: "選擇圖像生成模型",
-                                    enum: [
-                                        // 通義萬相模型
-                                        "wanx2.1-t2i-turbo",
-                                        "wanx2.1-t2i-plus",
-                                        "wanx2.0-t2i-turbo",
-                                        // 老張API模型
-                                        "sora_image",
-                                        "gpt-4o-image",
-                                        "gemini-2.5-flash-image-preview"
-                                    ],
-                                    default: "gemini-2.5-flash-image-preview",
-                                },
-                                negative_prompt: {
-                                    type: "string",
-                                    description: "反向提示詞，僅通義萬相模型支援（可選，預設為「人物」）",
-                                    default: "人物",
-                                },
-                                size: {
-                                    type: "string",
-                                    description: "輸出圖像的分辨率，僅通義萬相模型支援（可選，預設為 1024*1024）",
-                                    default: "1024*1024",
-                                },
-                                n: {
-                                    type: "number",
-                                    description: "生成圖片的數量，僅通義萬相模型支援（可選，範圍 1-4，預設為 1）",
-                                    minimum: 1,
-                                    maximum: 4,
-                                    default: 1,
-                                },
-                                seed: {
-                                    type: "number",
-                                    description: "隨機數種子，僅通義萬相模型支援（可選）",
-                                    minimum: 0,
-                                    maximum: 2147483647,
-                                },
-                                prompt_extend: {
-                                    type: "boolean",
-                                    description: "是否開啟 prompt 智能改寫，僅通義萬相模型支援（可選，預設為 true）",
-                                    default: true,
-                                },
-                                watermark: {
-                                    type: "boolean",
-                                    description: "是否添加水印標識，僅通義萬相模型支援（可選，預設為 false）",
-                                    default: false,
-                                },
-                                output_format: {
-                                    type: "string",
-                                    description: "輸出圖片格式（可選，預設為 jpg）",
-                                    enum: ["jpg", "jpeg", "png", "webp"],
-                                    default: "jpg",
-                                },
-                                system_prompt: {
-                                    type: "string",
-                                    description: "系統提示詞，僅老張API模型支援（可選）",
-                                },
-                                aspect_ratio: {
-                                    type: "string",
-                                    description: "圖片比例，僅老張API模型支援（可選），格式如：3:2, 16:9, 1:1",
-                                },
-                                max_wait_minutes: {
-                                    type: "number",
-                                    description: "最大等待時間（分鐘，可選，通義萬相預設5分鐘，老張API即時回應）",
-                                    minimum: 1,
-                                    maximum: 10,
-                                    default: 5,
-                                },
-                            },
-                            required: ["prompt"],
-                        },
-                    },
-                    {
-                        name: "tongyi_wanxiang_generate_video",
-                        description: "使用通義萬相圖生視頻完整流程（創建任務 + 等待完成 + 返回視頻URL）",
-                        inputSchema: {
-                            type: "object",
-                            properties: {
-                                img_url: {
-                                    type: "string",
-                                    description: "首幀圖像的 URL（必需，需為公網可訪問地址）",
-                                },
-                                prompt: {
-                                    type: "string",
-                                    description: "文本提示詞，支持中英文，長度不超過800字符（可選）",
-                                },
-                                model: {
-                                    type: "string",
-                                    description: "模型名稱 - turbo生成快(3-5分鐘)，plus品質高(7-10分鐘)",
-                                    enum: ["wanx2.1-i2v-turbo", "wanx2.1-i2v-plus"],
-                                    default: "wanx2.1-i2v-turbo",
-                                },
-                                template: {
-                                    type: "string",
-                                    description: "視頻特效模板（可選）：squish（解壓捏捏）、flying（魔法懸浮）、carousel（時光木馬）",
-                                    enum: ["squish", "flying", "carousel"],
-                                },
-                                resolution: {
-                                    type: "string",
-                                    description: "視頻分辨率檔位",
-                                    enum: ["480P", "720P"],
-                                    default: "720P",
-                                },
-                                duration: {
-                                    type: "number",
-                                    description: "視頻時長（秒）- turbo模型支持3-5秒，plus模型固定5秒",
-                                    minimum: 3,
-                                    maximum: 5,
-                                    default: 5,
-                                },
-                                prompt_extend: {
-                                    type: "boolean",
-                                    description: "是否開啟 prompt 智能改寫",
-                                    default: true,
-                                },
-                                seed: {
-                                    type: "number",
-                                    description: "隨機數種子",
-                                    minimum: 0,
-                                    maximum: 2147483647,
-                                },
-                                max_wait_minutes: {
-                                    type: "number",
-                                    description: "最大等待時間（分鐘，預設15分鐘）",
-                                    minimum: 5,
-                                    maximum: 30,
-                                    default: 15,
-                                },
-                            },
-                            required: ["img_url"],
-                        },
-                    },
-                    {
-                        name: "chat_completion",
-                        description: "使用 AI 進行對話",
-                        inputSchema: {
-                            type: "object",
-                            properties: {
-                                message: {
-                                    type: "string",
-                                    description: "用戶訊息",
-                                },
-                                system_prompt: {
-                                    type: "string",
-                                    description: "系統提示詞（可選，如未提供將使用預設值）",
-                                },
-                                model: {
-                                    type: "string",
-                                    description: "使用的模型名稱",
-                                    default: "gpt-4",
-                                },
-                            },
-                            required: ["message"],
-                        },
-                    },
-                    {
-                        name: "get_usage_guide",
-                        description: "獲取工具使用指南，包含所有可用模型的詳細說明",
-                        inputSchema: {
-                            type: "object",
-                            properties: {},
-                            required: [],
-                        },
-                    },
-                ],
-            };
-        });
-        // 處理工具調用
-        this.server.setRequestHandler(CallToolRequestSchema, async (request) => {
-            const { name, arguments: args } = request.params;
-            try {
-                switch (name) {
-                    case "generate_image":
-                        return await this.handleUnifiedImageGeneration(args);
-                    case "tongyi_wanxiang_generate_video":
-                        return await this.handleTongyiWanxiangGenerateVideo(args);
-                    case "chat_completion":
-                        return await this.handleChatCompletion(args);
-                    case "get_usage_guide":
-                        return await this.handleUsageGuide(args);
-                    // 保持向後兼容
-                    case "tongyi_wanxiang_generate_image":
-                        return await this.handleTongyiWanxiangGenerateImage(args);
-                    case "tongyi_wanxiang_create_task":
-                    case "dashscope_create_task":
-                        return await this.handleTongyiWanxiangCreateTask(args);
-                    case "tongyi_wanxiang_query_task":
-                    case "dashscope_query_task":
-                        return await this.handleTongyiWanxiangQueryTask(args);
-                    case "dashscope_generate_image":
-                        return await this.handleTongyiWanxiangGenerateImage(args);
-                    case "tongyi_wanxiang_create_video_task":
-                        return await this.handleTongyiWanxiangCreateVideoTask(args);
-                    default:
-                        throw new McpError(ErrorCode.MethodNotFound, `未知的工具: ${name}`);
-                }
-            }
-            catch (error) {
-                if (error instanceof McpError) {
-                    throw error;
-                }
-                throw new McpError(ErrorCode.InternalError, `工具執行錯誤: ${error instanceof Error ? error.message : String(error)}`);
-            }
+}
+async function queryDashScopeTask(taskId) {
+    if (!dashScopeApiKey) {
+        throw new Error("ALI_API_KEY 或 DASHSCOPE_API_KEY 環境變數未設定");
+    }
+    try {
+        const response = await axios.get(`https://dashscope.aliyuncs.com/api/v1/tasks/${taskId}`, {
+            headers: { Authorization: `Bearer ${dashScopeApiKey}` },
         });
+        return response.data;
     }
-    async handleImageGeneration(args) {
-        const { prompt, system_prompt, model = "gpt-4o-image", output_format = "jpg", n = 1, aspect_ratio } = args;
-        if (!prompt) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 prompt 參數");
-        }
-        // 模型資訊
-        const modelInfo = {
-            'sora_image': { name: 'Sora圖像生成', provider: '老張API', feature: '基於Sora技術的圖像生成，$0.01/次' },
-            'gpt-4o-image': { name: 'GPT-4o圖像生成', provider: '老張API', feature: 'GPT-4o視覺模型，$0.01/次' },
-            'gemini-2.5-flash-image-preview': { name: 'Gemini 2.5 Flash圖像生成', provider: '老張API', feature: 'Google Gemini 2.5 Flash預覽版，高速生成，$0.01/次' }
-        };
-        const modelDesc = modelInfo[model];
-        // 根據老張API文檔格式，構建正確的請求
-        let finalPrompt = prompt;
-        // 如果有比例要求，添加到提示詞末尾（如：【3:2】）
-        if (aspect_ratio) {
-            finalPrompt = `${prompt}【${aspect_ratio}】`;
-        }
-        const messages = [
-            {
-                role: "user",
-                content: [
-                    {
-                        type: "text",
-                        text: finalPrompt
-                    }
-                ]
-            }
-        ];
-        const requestData = {
-            model,
-            n: Math.min(Math.max(n, 1), 4), // 限制在1-4之間
-            messages
-        };
-        try {
-            const response = await axios.post(`${this.baseUrl}/chat/completions`, requestData, {
-                headers: {
-                    "Content-Type": "application/json",
-                    "Authorization": `Bearer ${this.apiKey}`,
-                },
-            });
-            const result = response.data;
-            const generatedContent = result.choices?.[0]?.message?.content || "未生成內容";
-            // 使用新的圖片處理邏輯，同時支援 URL 和 base64
-            const imageData = await this.processImageContent(generatedContent, prompt, output_format);
-            let responseText = `${modelDesc.name}圖像生成完成！\n\n`;
-            responseText += `🎨 使用模型: ${modelDesc.name}\n`;
-            responseText += `💰 費用: ${modelDesc.feature}\n`;
-            responseText += `📝 提示詞: ${finalPrompt}\n`;
-            responseText += `🖼️ 生成數量: ${n}張\n`;
-            if (system_prompt) {
-                responseText += `🔧 系統提示: ${system_prompt}\n`;
-            }
-            // 優先顯示 base64 圖片的保存結果
-            if (imageData.savedImages.length > 0) {
-                responseText += `\n💾 圖片已保存到 ${this.formatSaveDirectoryForDisplay()}:\n`;
-                imageData.savedImages.forEach((filePath, index) => {
-                    const displayPath = this.formatFilePathForDisplay(filePath);
-                    const filename = path.basename(filePath);
-                    responseText += `${index + 1}. ${filename}\n   完整路徑: ${displayPath}\n`;
-                });
-            }
-            // 如果有 URL 圖片，也顯示
-            if (imageData.imageUrls.length > 0) {
-                responseText += `\n🔗 圖片URLs:\n`;
-                imageData.imageUrls.forEach((url, index) => {
-                    responseText += `${index + 1}. ${url}\n`;
-                });
-            }
-            // 如果沒有找到任何圖片（URL 或 base64），顯示原始回應
-            if (!imageData.hasBase64 && imageData.imageUrls.length === 0) {
-                responseText += `\n📄 API回應內容:\n${generatedContent}`;
-            }
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: responseText,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            if (axios.isAxiosError(error)) {
-                const axiosError = error;
-                const errorMessage = axiosError.response?.data?.error?.message || axiosError.message;
-                // 根據文檔建議，如果sora_image失敗，建議切換到其他模型
-                let suggestionText = "";
-                if (model === "sora_image") {
-                    suggestionText = "\n💡 提示：如果sora_image持續失敗，建議切換到gpt-4o-image或gemini-2.5-flash-image-preview模型重試";
-                }
-                throw new McpError(ErrorCode.InternalError, `老張API請求失敗: ${errorMessage}${suggestionText}`);
-            }
-            throw new McpError(ErrorCode.InternalError, `未知錯誤: ${error instanceof Error ? error.message : String(error)}`);
+    catch (error) {
+        if (axios.isAxiosError(error)) {
+            const errorMessage = error.response?.data?.message || error.message;
+            throw new Error(`阿里雲 DashScope 查詢任務失敗: ${errorMessage}`);
         }
+        throw new Error(`查詢任務時發生未知錯誤: ${error}`);
     }
-    async handleTongyiWanxiangCreateTask(args) {
-        const { prompt, model, negative_prompt, size, n, seed, prompt_extend, watermark } = args;
-        if (!prompt) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 prompt 參數");
-        }
-        try {
-            const response = await this.createDashScopeTask(prompt, negative_prompt, {
-                model,
-                size,
-                n,
-                seed,
-                promptExtend: prompt_extend,
-                watermark,
-            });
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: `通義萬相文生圖任務創建成功！\n\n任務ID: ${response.output.task_id}\n任務狀態: ${response.output.task_status}\n使用模型: ${model || 'wanx2.1-t2i-turbo'}\n請求ID: ${response.request_id}\n\n請使用 tongyi_wanxiang_query_task 工具查詢任務結果。`,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            throw new McpError(ErrorCode.InternalError, `創建任務失敗: ${error instanceof Error ? error.message : String(error)}`);
-        }
+}
+async function generateImageWithDashScope(prompt, negativePrompt, options) {
+    const createResponse = await createDashScopeTask(prompt, negativePrompt, options);
+    const taskId = createResponse.output.task_id;
+    if (!taskId)
+        throw new Error("創建任務失敗，未獲取到任務ID");
+    const maxWaitMinutes = options?.maxWaitMinutes || 5;
+    const maxAttempts = maxWaitMinutes * 6;
+    let attempts = 0;
+    while (attempts < maxAttempts) {
+        const queryResponse = await queryDashScopeTask(taskId);
+        const status = queryResponse.output.task_status;
+        if (status === "SUCCEEDED")
+            return queryResponse.output.results || [];
+        if (status === "FAILED")
+            throw new Error(`圖片生成失敗: ${queryResponse.output.message || "未知錯誤"}`);
+        if (status === "CANCELED")
+            throw new Error("任務已被取消");
+        await new Promise((resolve) => setTimeout(resolve, 10000));
+        attempts++;
     }
-    async handleTongyiWanxiangQueryTask(args) {
-        const { task_id } = args;
-        if (!task_id) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 task_id 參數");
-        }
-        try {
-            const response = await this.queryDashScopeTask(task_id);
-            const { output } = response;
-            let responseText = `通義萬相文生圖任務查詢結果：\n\n`;
-            responseText += `任務ID: ${output.task_id}\n`;
-            responseText += `任務狀態: ${output.task_status}\n`;
-            responseText += `請求ID: ${response.request_id}\n`;
-            if (output.submit_time) {
-                responseText += `提交時間: ${output.submit_time}\n`;
-            }
-            if (output.scheduled_time) {
-                responseText += `開始時間: ${output.scheduled_time}\n`;
-            }
-            if (output.end_time) {
-                responseText += `完成時間: ${output.end_time}\n`;
-            }
-            if (output.task_status === "SUCCEEDED" && output.results) {
-                responseText += `\n✅ 任務完成！生成了 ${output.results.length} 張圖片：\n`;
-                output.results.forEach((result, index) => {
-                    responseText += `\n圖片 ${index + 1}:\n`;
-                    responseText += `原始提示詞: ${result.orig_prompt}\n`;
-                    if (result.actual_prompt) {
-                        responseText += `優化後提示詞: ${result.actual_prompt}\n`;
-                    }
-                    responseText += `圖片URL: ${result.url}\n`;
-                });
-                if (output.task_metrics) {
-                    responseText += `\n📊 任務統計:\n`;
-                    responseText += `總計: ${output.task_metrics.TOTAL}\n`;
-                    responseText += `成功: ${output.task_metrics.SUCCEEDED}\n`;
-                    responseText += `失敗: ${output.task_metrics.FAILED}\n`;
-                }
-            }
-            else if (output.task_status === "FAILED") {
-                responseText += `\n❌ 任務失敗\n`;
-                if (output.message) {
-                    responseText += `錯誤訊息: ${output.message}\n`;
-                }
-            }
-            else if (output.task_status === "PENDING") {
-                responseText += `\n⏳ 任務排隊中，請稍後再次查詢\n`;
-            }
-            else if (output.task_status === "RUNNING") {
-                responseText += `\n🔄 任務處理中，請稍後再次查詢\n`;
-            }
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: responseText,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            throw new McpError(ErrorCode.InternalError, `查詢任務失敗: ${error instanceof Error ? error.message : String(error)}`);
-        }
+    throw new Error(`圖片生成超時，等待時間超過 ${maxWaitMinutes} 分鐘`);
+}
+async function createVideoGenerationTask(imgUrl, prompt, options) {
+    if (!dashScopeApiKey) {
+        throw new Error("ALI_API_KEY 或 DASHSCOPE_API_KEY 環境變數未設定");
     }
-    async handleTongyiWanxiangGenerateImage(args) {
-        const { prompt, model, negative_prompt, size, n, seed, prompt_extend, watermark, output_format, max_wait_minutes } = args;
-        if (!prompt) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 prompt 參數");
-        }
-        // 模型資訊
-        const modelInfo = {
-            'wanx2.1-t2i-turbo': { name: '通義萬相2.1-Turbo', price: '0.14元/張', feature: '生成速度更快' },
-            'wanx2.1-t2i-plus': { name: '通義萬相2.1-Plus', price: '0.20元/張', feature: '圖像細節更豐富' },
-            'wanx2.0-t2i-turbo': { name: '通義萬相2.0-Turbo', price: '0.04元/張', feature: '質感人像與創意設計' }
-        };
-        const selectedModel = model || 'wanx2.1-t2i-turbo';
-        const modelDesc = modelInfo[selectedModel];
-        try {
-            const results = await this.generateImageWithDashScope(prompt, negative_prompt, {
-                model: selectedModel,
-                size,
-                n,
-                seed,
-                promptExtend: prompt_extend,
-                watermark,
-                outputFormat: output_format,
-                maxWaitMinutes: max_wait_minutes,
-            });
-            let responseText = `通義萬相文生圖像生成完成！\n\n`;
-            responseText += `🎨 使用模型: ${modelDesc.name} (${modelDesc.feature})\n`;
-            responseText += `💰 計費: ${modelDesc.price}\n`;
-            responseText += `✅ 成功生成 ${results.length} 張圖片\n\n`;
-            results.forEach((result, index) => {
-                responseText += `圖片 ${index + 1}:\n`;
-                responseText += `原始提示詞: ${result.orig_prompt}\n`;
-                if (result.actual_prompt) {
-                    responseText += `優化後提示詞: ${result.actual_prompt}\n`;
-                }
-                responseText += `圖片URL: ${result.url}\n\n`;
-            });
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: responseText,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            throw new McpError(ErrorCode.InternalError, `生成圖片失敗: ${error instanceof Error ? error.message : String(error)}`);
-        }
+    const requestData = {
+        model: options?.model || "wanx2.1-i2v-turbo",
+        input: {
+            img_url: imgUrl,
+            prompt,
+            template: options?.template,
+        },
+        parameters: {
+            resolution: options?.resolution || "720P",
+            duration: options?.duration || 5,
+            prompt_extend: options?.promptExtend !== false,
+            seed: options?.seed,
+        },
+    };
+    try {
+        const response = await axios.post("https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis", requestData, {
+            headers: {
+                "Content-Type": "application/json",
+                Authorization: `Bearer ${dashScopeApiKey}`,
+                "X-DashScope-Async": "enable",
+            },
+        });
+        return response.data;
     }
-    async handleTongyiWanxiangCreateVideoTask(args) {
-        const { img_url, prompt, model, template, resolution, duration, prompt_extend, seed } = args;
-        if (!img_url) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 img_url 參數");
-        }
-        try {
-            const response = await this.createVideoGenerationTask(img_url, prompt, {
-                model,
-                template,
-                resolution,
-                duration,
-                promptExtend: prompt_extend,
-                seed,
-            });
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: `通義萬相圖生視頻任務創建成功！\n\n任務ID: ${response.output.task_id}\n任務狀態: ${response.output.task_status}\n使用模型: ${model || 'wanx2.1-i2v-turbo'}\n首幀圖像: ${img_url}\n請求ID: ${response.request_id}\n\n請使用 tongyi_wanxiang_query_task 工具查詢任務結果。`,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            throw new McpError(ErrorCode.InternalError, `創建視頻生成任務失敗: ${error instanceof Error ? error.message : String(error)}`);
+    catch (error) {
+        if (axios.isAxiosError(error)) {
+            const errorMessage = error.response?.data?.message || error.message;
+            throw new Error(`阿里雲 DashScope 視頻生成任務創建失敗: ${errorMessage}`);
         }
+        throw new Error(`創建視頻生成任務時發生未知錯誤: ${error}`);
     }
-    async handleTongyiWanxiangGenerateVideo(args) {
-        const { img_url, prompt, model, template, resolution, duration, prompt_extend, seed, max_wait_minutes } = args;
-        if (!img_url) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 img_url 參數");
-        }
-        // 模型資訊
-        const modelInfo = {
-            'wanx2.1-i2v-turbo': { name: '通義萬相2.1圖生視頻-Turbo', time: '3-5分鐘', feature: '生成速度快' },
-            'wanx2.1-i2v-plus': { name: '通義萬相2.1圖生視頻-Plus', time: '7-10分鐘', feature: '視頻品質高' }
-        };
-        const selectedModel = model || 'wanx2.1-i2v-turbo';
-        const modelDesc = modelInfo[selectedModel];
-        try {
-            const videoUrl = await this.generateVideoWithDashScope(img_url, prompt, {
-                model: selectedModel,
-                template,
-                resolution,
-                duration,
-                promptExtend: prompt_extend,
-                seed,
-                maxWaitMinutes: max_wait_minutes,
-            });
-            let responseText = `通義萬相圖生視頻生成完成！\n\n`;
-            responseText += `🎬 使用模型: ${modelDesc.name} (${modelDesc.feature})\n`;
-            responseText += `⏱️ 預計生成時間: ${modelDesc.time}\n`;
-            responseText += `🖼️ 首幀圖像: ${img_url}\n`;
-            responseText += `📐 分辨率: ${resolution || '720P'}\n`;
-            responseText += `⏰ 視頻時長: ${duration || 5}秒\n`;
-            if (prompt) {
-                responseText += `💭 提示詞: ${prompt}\n`;
-            }
-            if (template) {
-                responseText += `✨ 特效模板: ${template}\n`;
-            }
-            responseText += `\n✅ 視頻生成成功！\n`;
-            responseText += `🔗 視頻URL: ${videoUrl}\n`;
-            responseText += `\n⚠️ 視頻URL有效期24小時，請及時保存`;
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: responseText,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            throw new McpError(ErrorCode.InternalError, `視頻生成失敗: ${error instanceof Error ? error.message : String(error)}`);
-        }
+}
+async function generateVideoWithDashScope(imgUrl, prompt, options) {
+    const createResponse = await createVideoGenerationTask(imgUrl, prompt, options);
+    const taskId = createResponse.output.task_id;
+    if (!taskId)
+        throw new Error("創建視頻生成任務失敗，未獲取到任務ID");
+    const maxWaitMinutes = options?.maxWaitMinutes || 15;
+    const maxAttempts = maxWaitMinutes * 6;
+    let attempts = 0;
+    while (attempts < maxAttempts) {
+        const queryResponse = await queryDashScopeTask(taskId);
+        const status = queryResponse.output.task_status;
+        if (status === "SUCCEEDED") {
+            const videoUrl = queryResponse.output.video_url;
+            if (!videoUrl)
+                throw new Error("任務完成但未獲取到視頻URL");
+            return videoUrl;
+        }
+        if (status === "FAILED")
+            throw new Error(`視頻生成失敗: ${queryResponse.output.message || "未知錯誤"}`);
+        if (status === "CANCELED")
+            throw new Error("視頻生成任務已被取消");
+        await new Promise((resolve) => setTimeout(resolve, 10000));
+        attempts++;
     }
-    async handleChatCompletion(args) {
-        const { message, system_prompt = "You are a helpful assistant.", model = "gpt-4" } = args;
-        if (!message) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 message 參數");
-        }
-        const messages = [
+    throw new Error(`視頻生成超時，等待時間超過 ${maxWaitMinutes} 分鐘`);
+}
+// ===== MCP Server =====
+const TONGYI_MODELS = ["wanx2.1-t2i-turbo", "wanx2.1-t2i-plus", "wanx2.0-t2i-turbo"];
+const LAOZHANG_MODELS = ["sora_image", "gpt-4o-image", "gemini-2.5-flash-image-preview"];
+const ALL_IMAGE_MODELS = [...TONGYI_MODELS, ...LAOZHANG_MODELS];
+const server = new McpServer({ name: "ai-image-chat-mcp-server", version: packageJson.version }, { capabilities: { logging: {} } });
+// ----- generate_image -----
+server.tool("generate_image", "統一圖像生成工具，支援通義萬相和老張API多種模型。預設使用 sora_image。", {
+    prompt: z.string().describe("圖像生成的提示詞"),
+    model: z.enum(ALL_IMAGE_MODELS).default("sora_image").describe("選擇圖像生成模型（預設: sora_image）"),
+    negative_prompt: z.string().default("人物").describe("反向提示詞，僅通義萬相模型支援").optional(),
+    size: z.string().default("1024*1024").describe("輸出圖像的分辨率，僅通義萬相模型支援").optional(),
+    n: z.number().min(1).max(4).default(1).describe("生成圖片數量（1-4）").optional(),
+    seed: z.number().min(0).max(2147483647).describe("隨機數種子，僅通義萬相模型支援").optional(),
+    prompt_extend: z.boolean().default(true).describe("是否開啟 prompt 智能改寫，僅通義萬相模型支援").optional(),
+    watermark: z.boolean().default(false).describe("是否添加水印標識，僅通義萬相模型支援").optional(),
+    output_format: z.enum(["jpg", "jpeg", "png", "webp"]).default("jpg").describe("輸出圖片格式").optional(),
+    system_prompt: z.string().describe("系統提示詞，僅老張API模型支援").optional(),
+    aspect_ratio: z.string().describe("圖片比例，僅老張API模型支援，格式如：3:2, 16:9, 1:1").optional(),
+    max_wait_minutes: z.number().min(1).max(10).default(5).describe("最大等待時間（分鐘，通義萬相預設5分鐘，老張API即時回應）").optional(),
+}, async (args) => {
+    const { prompt, model = "sora_image", negative_prompt, size, n = 1, seed, prompt_extend, watermark, output_format = "jpg", system_prompt, aspect_ratio, max_wait_minutes, } = args;
+    if (TONGYI_MODELS.includes(model)) {
+        return handleTongyiImageGeneration({
+            prompt, model, negative_prompt, size, n, seed,
+            prompt_extend, watermark, output_format, max_wait_minutes,
+        });
+    }
+    return handleLaozhangImageGeneration({
+        prompt, model, system_prompt, output_format, n, aspect_ratio,
+    });
+});
+// ----- tongyi_wanxiang_generate_video -----
+server.tool("tongyi_wanxiang_generate_video", "使用通義萬相圖生視頻完整流程（創建任務 + 等待完成 + 返回視頻URL）", {
+    img_url: z.string().describe("首幀圖像的 URL（必需，需為公網可訪問地址）"),
+    prompt: z.string().describe("文本提示詞，支持中英文，長度不超過800字符").optional(),
+    model: z.enum(["wanx2.1-i2v-turbo", "wanx2.1-i2v-plus"]).default("wanx2.1-i2v-turbo").describe("模型名稱 - turbo生成快(3-5分鐘)，plus品質高(7-10分鐘)").optional(),
+    template: z.enum(["squish", "flying", "carousel"]).describe("視頻特效模板：squish（解壓捏捏）、flying（魔法懸浮）、carousel（時光木馬）").optional(),
+    resolution: z.enum(["480P", "720P"]).default("720P").describe("視頻分辨率檔位").optional(),
+    duration: z.number().min(3).max(5).default(5).describe("視頻時長（秒）- turbo模型支持3-5秒，plus模型固定5秒").optional(),
+    prompt_extend: z.boolean().default(true).describe("是否開啟 prompt 智能改寫").optional(),
+    seed: z.number().min(0).max(2147483647).describe("隨機數種子").optional(),
+    max_wait_minutes: z.number().min(5).max(30).default(15).describe("最大等待時間（分鐘，預設15分鐘）").optional(),
+}, async (args) => {
+    const { img_url, prompt, model = "wanx2.1-i2v-turbo", template, resolution = "720P", duration = 5, prompt_extend, seed, max_wait_minutes = 15, } = args;
+    const modelInfo = {
+        "wanx2.1-i2v-turbo": { name: "通義萬相2.1圖生視頻-Turbo", time: "3-5分鐘", feature: "生成速度快" },
+        "wanx2.1-i2v-plus": { name: "通義萬相2.1圖生視頻-Plus", time: "7-10分鐘", feature: "視頻品質高" },
+    };
+    const desc = modelInfo[model];
+    const videoUrl = await generateVideoWithDashScope(img_url, prompt, {
+        model, template, resolution, duration,
+        promptExtend: prompt_extend, seed, maxWaitMinutes: max_wait_minutes,
+    });
+    let text = `通義萬相圖生視頻生成完成！\n\n`;
+    text += `🎬 使用模型: ${desc.name} (${desc.feature})\n`;
+    text += `⏱️ 預計生成時間: ${desc.time}\n`;
+    text += `🖼️ 首幀圖像: ${img_url}\n`;
+    text += `📐 分辨率: ${resolution}\n`;
+    text += `⏰ 視頻時長: ${duration}秒\n`;
+    if (prompt)
+        text += `💭 提示詞: ${prompt}\n`;
+    if (template)
+        text += `✨ 特效模板: ${template}\n`;
+    text += `\n✅ 視頻生成成功！\n`;
+    text += `🔗 視頻URL: ${videoUrl}\n`;
+    text += `\n⚠️ 視頻URL有效期24小時，請及時保存`;
+    return { content: [{ type: "text", text }] };
+});
+// ----- chat_completion -----
+server.tool("chat_completion", "使用 AI 進行對話", {
+    message: z.string().describe("用戶訊息"),
+    system_prompt: z.string().default("You are a helpful assistant.").describe("系統提示詞").optional(),
+    model: z.string().default("gpt-4o").describe("使用的模型名稱（預設: gpt-4o）").optional(),
+}, async (args) => {
+    const { message, system_prompt = "You are a helpful assistant.", model = "gpt-4o" } = args;
+    const response = await axios.post(`${baseUrl}/chat/completions`, {
+        model,
+        messages: [
             { role: "system", content: system_prompt },
             { role: "user", content: message },
-        ];
-        const requestData = {
-            model,
-            messages,
-        };
-        try {
-            const response = await axios.post(`${this.baseUrl}/chat/completions`, requestData, {
-                headers: {
-                    "Content-Type": "application/json",
-                    "Authorization": `Bearer ${this.apiKey}`,
-                },
-            });
-            const result = response.data;
-            const assistantMessage = result.choices[0]?.message?.content || "未收到回覆";
-            return {
-                content: [
-                    {
-                        type: "text",
-                        text: `助手回覆:\n\n${assistantMessage}`,
-                    },
-                ],
-            };
-        }
-        catch (error) {
-            if (axios.isAxiosError(error)) {
-                const axiosError = error;
-                const errorMessage = axiosError.response?.data?.error?.message || axiosError.message;
-                throw new McpError(ErrorCode.InternalError, `API 請求失敗: ${errorMessage}`);
-            }
-            throw new McpError(ErrorCode.InternalError, `未知錯誤: ${error instanceof Error ? error.message : String(error)}`);
-        }
+        ],
+    }, {
+        headers: {
+            "Content-Type": "application/json",
+            Authorization: `Bearer ${apiKey}`,
+        },
+    });
+    const assistantMessage = response.data.choices[0]?.message?.content || "未收到回覆";
+    return { content: [{ type: "text", text: `助手回覆:\n\n${assistantMessage}` }] };
+});
+// ----- get_usage_guide -----
+server.tool("get_usage_guide", "獲取工具使用指南，包含所有可用模型的詳細說明", {}, async () => {
+    return { content: [{ type: "text", text: buildUsageGuide() }] };
+});
+// ===== 處理函式 =====
+async function handleTongyiImageGeneration(args) {
+    const modelInfo = {
+        "wanx2.1-t2i-turbo": { name: "通義萬相2.1-Turbo", price: "0.14元/張", feature: "生成速度更快" },
+        "wanx2.1-t2i-plus": { name: "通義萬相2.1-Plus", price: "0.20元/張", feature: "圖像細節更豐富" },
+        "wanx2.0-t2i-turbo": { name: "通義萬相2.0-Turbo", price: "0.04元/張", feature: "質感人像與創意設計" },
+    };
+    const desc = modelInfo[args.model];
+    const results = await generateImageWithDashScope(args.prompt, args.negative_prompt, {
+        model: args.model,
+        size: args.size,
+        n: args.n,
+        seed: args.seed,
+        promptExtend: args.prompt_extend,
+        watermark: args.watermark,
+        maxWaitMinutes: args.max_wait_minutes,
+    });
+    let text = `通義萬相文生圖像生成完成！\n\n`;
+    text += `🎨 使用模型: ${desc.name} (${desc.feature})\n`;
+    text += `💰 計費: ${desc.price}\n`;
+    text += `✅ 成功生成 ${results.length} 張圖片\n\n`;
+    for (const [i, result] of results.entries()) {
+        text += `圖片 ${i + 1}:\n`;
+        text += `原始提示詞: ${result.orig_prompt}\n`;
+        if (result.actual_prompt)
+            text += `優化後提示詞: ${result.actual_prompt}\n`;
+        text += `圖片URL: ${result.url}\n\n`;
     }
-    async handleUnifiedImageGeneration(args) {
-        const { prompt, model = "gemini-2.5-flash-image-preview", negative_prompt, size, n, seed, prompt_extend, watermark, output_format = "jpg", system_prompt, aspect_ratio, max_wait_minutes } = args;
-        if (!prompt) {
-            throw new McpError(ErrorCode.InvalidParams, "需要提供 prompt 參數");
-        }
-        // 判斷是通義萬相模型還是老張API模型
-        const tongyiModels = ["wanx2.1-t2i-turbo", "wanx2.1-t2i-plus", "wanx2.0-t2i-turbo"];
-        const laozhangModels = ["sora_image", "gpt-4o-image", "gemini-2.5-flash-image-preview"];
-        if (tongyiModels.includes(model)) {
-            // 使用通義萬相API
-            return await this.handleTongyiWanxiangGenerateImage({
-                prompt,
-                model,
-                negative_prompt,
-                size,
-                n,
-                seed,
-                prompt_extend,
-                watermark,
-                output_format,
-                max_wait_minutes
-            });
-        }
-        else if (laozhangModels.includes(model)) {
-            // 使用老張API
-            return await this.handleImageGeneration({
-                prompt,
-                model,
-                system_prompt,
-                output_format,
-                n,
-                aspect_ratio
-            });
+    return { content: [{ type: "text", text }] };
+}
+async function handleLaozhangImageGeneration(args) {
+    const { prompt, model, system_prompt, output_format = "jpg", n = 1, aspect_ratio } = args;
+    const modelInfo = {
+        sora_image: { name: "Sora圖像生成", feature: "基於Sora技術的圖像生成，$0.01/次" },
+        "gpt-4o-image": { name: "GPT-4o圖像生成", feature: "GPT-4o視覺模型，$0.01/次" },
+        "gemini-2.5-flash-image-preview": { name: "Gemini 2.5 Flash圖像生成", feature: "Google Gemini 2.5 Flash預覽版，高速生成，$0.01/次" },
+    };
+    const desc = modelInfo[model];
+    const finalPrompt = aspect_ratio ? `${prompt}【${aspect_ratio}】` : prompt;
+    const response = await axios.post(`${baseUrl}/chat/completions`, {
+        model,
+        n: Math.min(Math.max(n, 1), 4),
+        messages: [{ role: "user", content: [{ type: "text", text: finalPrompt }] }],
+    }, {
+        headers: {
+            "Content-Type": "application/json",
+            Authorization: `Bearer ${apiKey}`,
+        },
+    });
+    const generatedContent = response.data.choices?.[0]?.message?.content || "未生成內容";
+    const imageData = await processImageContent(generatedContent, prompt, output_format);
+    let text = `${desc.name}圖像生成完成！\n\n`;
+    text += `🎨 使用模型: ${desc.name}\n`;
+    text += `💰 費用: ${desc.feature}\n`;
+    text += `📝 提示詞: ${finalPrompt}\n`;
+    text += `🖼️ 生成數量: ${n}張\n`;
+    if (system_prompt)
+        text += `🔧 系統提示: ${system_prompt}\n`;
+    if (imageData.savedImages.length > 0) {
+        text += `\n💾 圖片已保存到 ${formatPathForDisplay(saveDirectory)}:\n`;
+        for (const [i, filePath] of imageData.savedImages.entries()) {
+            text += `${i + 1}. ${path.basename(filePath)}\n   完整路徑: ${formatPathForDisplay(filePath)}\n`;
         }
-        else {
-            throw new McpError(ErrorCode.InvalidParams, `不支援的模型: ${model}`);
+    }
+    if (imageData.imageUrls.length > 0) {
+        text += `\n🔗 圖片URLs:\n`;
+        for (const [i, url] of imageData.imageUrls.entries()) {
+            text += `${i + 1}. ${url}\n`;
         }
     }
-    async handleUsageGuide(args) {
-        const guideText = `
+    if (!imageData.hasBase64 && imageData.imageUrls.length === 0) {
+        text += `\n📄 API回應內容:\n${generatedContent}`;
+    }
+    return { content: [{ type: "text", text }] };
+}
+// ===== 使用指南 =====
+function buildUsageGuide() {
+    return `
 # AI Image Chat MCP 工具使用指南
 
 **版本**: v${packageJson.version}
-**最後更新**: ${new Date().toLocaleDateString('zh-TW')}
+**最後更新**: ${new Date().toLocaleDateString("zh-TW")}
 **支援模型**: 6 種圖像生成模型 + 2 種視頻生成模型
 
 ## 🎨 統一圖像生成工具
@@ -1006,50 +438,52 @@ class AIImageChatMCPServer {
 
 **可用模型**:
 
+### 老張API模型（即時回應，$0.01/次）
+
+1. **sora_image** ⭐ (預設)
+   - 提供商: 老張API
+   - 特點: 基於Sora技術的圖像生成
+   - 生成時間: 即時回應
+   - 適合: 快速圖像生成、創意探索、首選模型
+
+2. **gpt-4o-image**
+   - 提供商: 老張API
+   - 特點: GPT-4o視覺模型
+   - 生成時間: 即時回應
+   - 適合: 智能圖像理解和生成
+
+3. **gemini-2.5-flash-image-preview**
+   - 提供商: 老張API
+   - 特點: Google Gemini 2.5 Flash預覽版，高速生成
+   - 生成時間: 即時回應
+   - 適合: 高速生成、高效批量生成
+
 ### 通義萬相模型（阿里雲DashScope）
-1. **wanx2.1-t2i-turbo**
+
+4. **wanx2.1-t2i-turbo**
    - 提供商: 阿里雲
    - 價格: 0.14元/張
    - 特點: 生成速度更快
    - 生成時間: 1-3分鐘
    - 適合: 快速原型和測試
 
-2. **wanx2.1-t2i-plus**
+5. **wanx2.1-t2i-plus**
    - 提供商: 阿里雲
    - 價格: 0.20元/張
    - 特點: 圖像細節更豐富
    - 生成時間: 1-3分鐘
    - 適合: 高品質作品和商業用途
 
-3. **wanx2.0-t2i-turbo**
+6. **wanx2.0-t2i-turbo**
    - 提供商: 阿里雲
    - 價格: 0.04元/張
    - 特點: 質感人像與創意設計
    - 生成時間: 1-3分鐘
    - 適合: 成本控制和批量生成
 
-### 老張API模型（即時回應，$0.01/次）
-4. **gemini-2.5-flash-image-preview** ⭐ (預設)
-   - 提供商: 老張API
-   - 特點: Google Gemini 2.5 Flash預覽版，高速生成
-   - 生成時間: 即時回應
-   - 適合: 快速原型設計、高效批量生成、首選模型
-
-5. **sora_image**
-   - 提供商: 老張API
-   - 特點: 基於Sora技術的圖像生成
-   - 生成時間: 即時回應
-   - 適合: 快速圖像生成、創意探索
-
-6. **gpt-4o-image**
-   - 提供商: 老張API
-   - 特點: GPT-4o視覺模型
-   - 生成時間: 即時回應
-   - 適合: 智能圖像理解和生成
-
 **主要參數**:
 - \`prompt\`: 圖像生成提示詞 (必需)
-- \`model\`: 模型選擇 (可選，預設: "gemini-2.5-flash-image-preview")
+- \`model\`: 模型選擇 (可選，預設: "sora_image")
 
 **通義萬相模型專用參數**:
 - \`negative_prompt\`: 反向提示詞 (可選，預設: "人物")
@@ -1062,6 +496,7 @@ class AIImageChatMCPServer {
 
 **老張API模型專用參數**:
 - \`system_prompt\`: 系統提示詞 (可選)
+- \`aspect_ratio\`: 圖片比例 (可選，格式如 "3:2", "16:9")
 - \`output_format\`: 圖片格式 (可選，預設: "jpg")
 
 ---
@@ -1072,39 +507,10 @@ class AIImageChatMCPServer {
 **功能**: 使用通義萬相圖生視頻完整流程（創建任務 + 等待完成 + 返回視頻URL）
 
 **可用模型**:
-1. **wanx2.1-i2v-turbo** (預設)
-   - 生成時間: 3-5分鐘
-   - 特點: 生成速度快
-   - 支援時長: 3-5秒
-   - 支援分辨率: 480P, 720P
+1. **wanx2.1-i2v-turbo** (預設) - 生成快(3-5分鐘)，支援3-5秒
+2. **wanx2.1-i2v-plus** - 品質高(7-10分鐘)，固定5秒
 
-2. **wanx2.1-i2v-plus**
-   - 生成時間: 7-10分鐘
-   - 特點: 視頻品質高
-   - 支援時長: 固定5秒
-   - 支援分辨率: 僅720P
-
-**主要參數**:
-- \`img_url\`: 首幀圖像URL (必需，需為公網可訪問地址)
-- \`prompt\`: 文本提示詞 (可選，長度不超過800字符)
-- \`model\`: 模型選擇 (可選)
-- \`template\`: 視頻特效模板 (可選)
-- \`resolution\`: 分辨率檔位 (可選，預設: "720P")
-- \`duration\`: 視頻時長 (可選，預設: 5秒)
-- \`prompt_extend\`: 智能改寫 (可選，預設: true)
-- \`seed\`: 隨機種子 (可選)
-- \`max_wait_minutes\`: 最大等待時間 (可選，預設: 15分鐘)
-
-**視頻特效模板**:
-- \`squish\`: 解壓捏捏效果
-- \`flying\`: 魔法懸浮效果
-- \`carousel\`: 時光木馬效果
-
-**圖像要求**:
-- 格式: JPEG、JPG、PNG、BMP、WEBP
-- 尺寸: 360-2000像素（寬度和高度）
-- 檔案大小: 不超過10MB
-- URL: 必須為公網可訪問地址
+**視頻特效模板**: squish（解壓捏捏）、flying（魔法懸浮）、carousel（時光木馬）
 
 ---
 
@@ -1112,276 +518,107 @@ class AIImageChatMCPServer {
 
 ### chat_completion
 **功能**: 使用 AI 進行對話
-
-**主要參數**:
 - \`message\`: 用戶訊息 (必需)
 - \`system_prompt\`: 系統提示詞 (可選)
-- \`model\`: 模型名稱 (可選，預設: "gpt-4")
+- \`model\`: 模型名稱 (可選，預設: "gpt-4o")
 
 ---
 
 ## 📋 使用範例
 
-### 📋 圖像生成模型詳細調用範例
-
-#### 1. gemini-2.5-flash-image-preview (預設) - 高速生成
-**特點**: Gemini 2.5 Flash預覽版 | **價格**: $0.01/次 | **時間**: 即時回應
-\`\`\`json
-{
-  "name": "generate_image",
-  "arguments": {
-    "prompt": "一隻可愛的橘貓在陽光下睡覺",
-    "model": "gemini-2.5-flash-image-preview",
-    "aspect_ratio": "16:9",
-    "n": 2
-  }
-}
-\`\`\`
-
-#### 2. wanx2.1-t2i-turbo - 速度優先
-**特點**: 生成速度快 | **價格**: 0.14元/張 | **時間**: 1-3分鐘
-\`\`\`json
-{
-  "name": "generate_image",
-  "arguments": {
-    "prompt": "一隻可愛的橘貓在陽光下睡覺",
-    "model": "wanx2.1-t2i-turbo",
-    "size": "1024*1024",
-    "n": 1,
-    "negative_prompt": "人物，文字，低質量",
-    "prompt_extend": true,
-    "watermark": false,
-    "max_wait_minutes": 5
-  }
-}
-\`\`\`
-
-#### 2. wanx2.1-t2i-turbo - 速度優先
-**特點**: 生成速度快 | **價格**: 0.14元/張 | **時間**: 1-3分鐘
-\`\`\`json
-{
-  "name": "generate_image",
-  "arguments": {
-    "prompt": "一隻可愛的橘貓在陽光下睡覺",
-    "model": "wanx2.1-t2i-turbo",
-    "size": "1024*1024",
-    "n": 1,
-    "negative_prompt": "人物，文字，低質量",
-    "prompt_extend": true,
-    "watermark": false,
-    "max_wait_minutes": 5
-  }
-}
-\`\`\`
-
-#### 3. wanx2.1-t2i-plus - 品質優先  
-**特點**: 圖像細節豐富 | **價格**: 0.20元/張 | **時間**: 1-3分鐘
-\`\`\`json
-{
-  "name": "generate_image",
-  "arguments": {
-    "prompt": "精美的日式庭院，櫻花飄落，寧靜祥和",
-    "model": "wanx2.1-t2i-plus",
-    "size": "1024*1024",
-    "n": 2,
-    "negative_prompt": "人物，文字，模糊",
-    "seed": 12345,
-    "prompt_extend": true,
-    "output_format": "jpg"
-  }
-}
-\`\`\`
-
-#### 4. wanx2.0-t2i-turbo - 性價比首選
-**特點**: 質感人像與創意設計 | **價格**: 0.04元/張 | **時間**: 1-3分鐘  
-\`\`\`json
-{
-  "name": "generate_image",
-  "arguments": {
-    "prompt": "時尚的年輕女性肖像，現代都市背景",
-    "model": "wanx2.0-t2i-turbo",
-    "size": "1024*1024",
-    "n": 4,
-    "negative_prompt": "醜陋，變形，低分辨率",
-    "max_wait_minutes": 3
-  }
-}
-\`\`\`
-
-#### 5. sora_image - 即時生成
-**特點**: 基於Sora技術 | **價格**: $0.01/次 | **時間**: 即時回應
+#### 1. sora_image (預設) - 即時生成
 \`\`\`json
 {
   "name": "generate_image",
   "arguments": {
     "prompt": "夢幻般的未來城市，科技感十足，霓虹燈光",
-    "model": "sora_image",
-    "n": 4,
     "aspect_ratio": "16:9"
   }
 }
 \`\`\`
 
-#### 6. gpt-4o-image - 智能生成
-**特點**: GPT-4o視覺模型 | **價格**: $0.01/次 | **時間**: 即時回應
+#### 2. gpt-4o-image - 智能生成
 \`\`\`json
 {
   "name": "generate_image",
   "arguments": {
     "prompt": "可愛的卡通動物們在森林裡開派對",
     "model": "gpt-4o-image",
-    "n": 2,
     "aspect_ratio": "3:2",
     "system_prompt": "Create a colorful, family-friendly cartoon style image"
   }
 }
 \`\`\`
 
-**特點**: Gemini 2.5 Flash預覽版 | **價格**: $0.01/次 | **時間**: 即時回應
+#### 3. gemini-2.5-flash-image-preview - 高速生成
 \`\`\`json
 {
   "name": "generate_image",
   "arguments": {
     "prompt": "賽博朋克風格的東京街道，雨夜霓虹",
     "model": "gemini-2.5-flash-image-preview",
-    "n": 3,
     "aspect_ratio": "16:9"
   }
 }
 \`\`\`
 
-#### 6. 老張API比例控制範例
-**展示不同比例效果**
+#### 4. wanx2.1-t2i-plus - 品質優先
 \`\`\`json
 {
   "name": "generate_image",
   "arguments": {
-    "prompt": "一隻貓咪坐在窗台上看風景",
-    "model": "sora_image",
-    "aspect_ratio": "5:3",
-    "n": 1
+    "prompt": "精美的日式庭院，櫻花飄落，寧靜祥和",
+    "model": "wanx2.1-t2i-plus",
+    "size": "1024*1024",
+    "n": 2,
+    "negative_prompt": "人物，文字，模糊"
   }
 }
 \`\`\`
 
-### 📋 詳細參數說明
-
-#### 🔧 通用參數 (所有模型)
-- \`**prompt**\` (必需): 圖像生成的描述詞
-- \`**model**\` (可選): 選擇生成模型，預設 "gemini-2.5-flash-image-preview"
-- \`**output_format**\` (可選): 圖片格式 ["jpg", "jpeg", "png", "webp"]，預設 "jpg"
-
-#### ⚙️ 通義萬相專用參數 (wanx2.x 模型)
-- \`**negative_prompt**\` (可選): 反向提示詞，描述不希望出現的內容，預設 "人物"
-- \`**size**\` (可選): 圖像尺寸，預設 "1024*1024"
-  - 支援: "512*512", "768*768", "1024*1024", "1280*720", "720*1280"
-- \`**n**\` (可選): 生成圖片數量，範圍 1-4，預設 1
-- \`**seed**\` (可選): 隨機種子 (0-2147483647)，用於重現相同結果
-- \`**prompt_extend**\` (可選): 智能改寫提示詞，預設 true
-- \`**watermark**\` (可選): 添加水印標識，預設 false
-- \`**max_wait_minutes**\` (可選): 最大等待時間 (1-10分鐘)，預設 5分鐘
-
-#### 🤖 老張API專用參數 (sora_image, gpt-4o-image, gemini-2.5-flash-image-preview)
-- \`**system_prompt**\` (可選): 系統提示詞，用於指導AI生成風格
-- \`**aspect_ratio**\` (可選): 圖片比例，格式如 "3:2", "16:9", "1:1" (會添加到提示詞末尾)
-- \`**n**\` (可選): 生成圖片數量，範圍 1-4，預設 1 (注意：老張API按次計費，每次$0.01)
-- \`**max_wait_minutes**\` (可選): 最大等待時間，預設 5分鐘 (通常即時回應)
-
-### 💰 模型價格對比表
-
-| 模型 | 提供商 | 價格 | 生成時間 | 適用場景 |
-|------|---------|------|----------|----------|
-| **wanx2.0-t2i-turbo** | 阿里雲 | **0.04元/張** | 1-3分鐘 | 💰 批量生成、成本控制 |
-| **wanx2.1-t2i-turbo** | 阿里雲 | 0.14元/張 | 1-3分鐘 | ⚡ 快速原型、測試 |
-| **wanx2.1-t2i-plus** | 阿里雲 | 0.20元/張 | 1-3分鐘 | 🎨 高品質、商業用途 |
-| **sora_image** | 老張API | **$0.01/次** | 即時 | 🚀 即時生成、創意探索 |
-| **gpt-4o-image** | 老張API | **$0.01/次** | 即時 | 🤖 智能理解、複雜場景 |
-| **gemini-2.5-flash-image-preview** | 老張API | **$0.01/次** | 即時 | ⚡ 高速生成、高效處理 |
-
-### 💡 老張API使用說明
-- **計費方式**: 按次計費，每次調用 $0.01，不論生成幾張圖片
-- **比例控制**: 在提示詞末尾自動添加【比例】格式，如【3:2】
-- **重試機制**: 如果 sora_image 失敗，建議切換到 gpt-4o-image 或 gemini-2.5-flash-image-preview
-- **數量控制**: 使用 n 參數控制生成數量 (1-4張)
-- **格式特點**: 基於 ChatGPT PLUS 用戶的生圖請求模擬
-
-### 🎯 模型選擇建議
-
-**追求速度**: \`gemini-2.5-flash-image-preview\`, \`sora_image\`, \`gpt-4o-image\` (即時回應)
-**注重成本**: \`wanx2.0-t2i-turbo\` (最便宜，0.04元/張)  
-**平衡選擇**: \`wanx2.1-t2i-turbo\` (速度與品質平衡)
-**追求品質**: \`wanx2.1-t2i-plus\` (最高品質)
-**創意實驗**: \`sora_image\` (Sora技術，獨特風格)
-
-### 視頻生成範例
+#### 5. 視頻生成範例
 \`\`\`json
 {
   "name": "tongyi_wanxiang_generate_video",
   "arguments": {
     "img_url": "https://example.com/cat.jpg",
     "prompt": "小貓在草地上慢慢伸懶腰",
-    "model": "wanx2.1-i2v-turbo",
     "duration": 4,
     "resolution": "720P"
   }
 }
 \`\`\`
 
-### 特效視頻範例
-\`\`\`json
-{
-  "name": "tongyi_wanxiang_generate_video",
-  "arguments": {
-    "img_url": "https://example.com/object.jpg",
-    "template": "flying",
-    "resolution": "720P"
-  }
-}
-\`\`\`
-
 ---
 
-## ⚠️ 注意事項
+## 💰 模型價格對比表
 
-1. **圖像生成**: 通常需要1-3分鐘完成
-2. **視頻生成**: turbo模型3-5分鐘，plus模型7-10分鐘
-3. **URL有效期**: 生成的圖片和視頻URL有效期24小時
-4. **等待機制**: 工具會自動輪詢任務狀態直到完成
-5. **錯誤處理**: 任務失敗時會提供詳細錯誤信息
-6. **環境變數**: 需要設置 ALI_API_KEY 或 DASHSCOPE_API_KEY
+| 模型 | 提供商 | 價格 | 生成時間 | 適用場景 |
+|------|---------|------|----------|----------|
+| **sora_image** ⭐ | 老張API | **$0.01/次** | 即時 | 🚀 首選，即時生成 |
+| **gpt-4o-image** | 老張API | **$0.01/次** | 即時 | 🤖 智能理解、複雜場景 |
+| **gemini-2.5-flash-image-preview** | 老張API | **$0.01/次** | 即時 | ⚡ 高速生成 |
+| **wanx2.0-t2i-turbo** | 阿里雲 | **0.04元/張** | 1-3分鐘 | 💰 批量生成、成本控制 |
+| **wanx2.1-t2i-turbo** | 阿里雲 | 0.14元/張 | 1-3分鐘 | ⚡ 快速原型、測試 |
+| **wanx2.1-t2i-plus** | 阿里雲 | 0.20元/張 | 1-3分鐘 | 🎨 高品質、商業用途 |
 
 ---
 
 ## 🔧 環境配置
 
-確保設置以下環境變數：
-- \`ALI_API_KEY\` 或 \`DASHSCOPE_API_KEY\`: 阿里雲DashScope API密鑰（通義萬相模型需要）
-- \`AI_API_KEY\`: 老張API密鑰（老張API模型和chat_completion需要）
+- \`AI_API_KEY\`: 老張API密鑰（必需）
 - \`AI_API_BASE_URL\`: 老張API基礎URL（可選，預設: https://api.laozhang.ai/v1）
-
-**API密鑰配置說明**:
-- 使用通義萬相模型時，需要配置阿里雲API密鑰
-- 使用老張API模型時，需要配置老張API密鑰
-- 如果同時使用兩種模型，需要配置兩個API密鑰
-`;
-        return {
-            content: [
-                {
-                    type: "text",
-                    text: guideText.trim(),
-                },
-            ],
-        };
-    }
-    async run() {
-        const transport = new StdioServerTransport();
-        await this.server.connect(transport);
-        console.error(`AI Image Chat MCP 服務器已啟動 - v${packageJson.version}`);
-    }
+- \`ALI_API_KEY\` 或 \`DASHSCOPE_API_KEY\`: 阿里雲DashScope API密鑰（使用通義萬相時需要）
+- \`AI_IMAGE_SAVE_PATH\`: 圖片保存路徑（可選）
+`.trim();
+}
+// ===== 啟動 =====
+async function main() {
+    const transport = new StdioServerTransport();
+    await server.connect(transport);
+    console.error(`AI Image Chat MCP 服務器已啟動 - v${packageJson.version}`);
 }
-// 啟動服務器
-const server = new AIImageChatMCPServer();
-server.run().catch((error) => {
+main().catch((error) => {
     console.error("服務器啟動失敗:", error);
     process.exit(1);
 });